One read of Model Context Protocolmcp-20261020T200724Z
306 pages moved out of 357 read.
Pages moved
306
significant first
Pages read
357
in this capture
Captured
20:07 UTC
Corpus hash
077a51951f96
corpus-hash
What this read moved
1-25 of 306, page 1 of 13This capture is too large to show at once. Changes 1-25 of 306 are below, significant first; the rest are on the following screens.
community/antitrust Changed · +7 / -7 lines
from line 4
44
55**Effective: September 29, 2025**
66
7<Note>
8 This policy applies when participating in MCP meetings, Working Groups,
9 Interest Groups, and other collaborative forums where competitors may be
10 present. Most individual contributors working on code or documentation don't
11 need to worry about this in day-to-day work - it's primarily relevant for
12 group discussions about standards and specifications.
13</Note>
7> [!NOTE]
8>
9> This policy applies when participating in MCP meetings, Working Groups,
10> Interest Groups, and other collaborative forums where competitors may be
11> present. Most individual contributors working on code or documentation don't
12> need to worry about this in day-to-day work - it's primarily relevant for
13> group discussions about standards and specifications.
1414
1515## Introduction
1616
community/charter-template Changed · +3 / -3 lines
from line 8
88
99Sections marked **(WG only)** are required for Working Groups and optional for Interest Groups.
1010
11<Note>
12 Copy the markdown below into `docs/community/working-groups/<group-name>.mdx` (or `interest-groups/` for an IG) and replace the placeholder text.
13</Note>
11> [!NOTE]
12>
13> Copy the markdown below into `docs/community/working-groups/<group-name>.mdx` (or `interest-groups/` for an IG) and replace the placeholder text.
1414
1515***
1616
community/contributing Changed · +225 / -243 lines
from line 35
3535git --version # Any recent version
3636```
3737
38<Note>
39 These commands work the same on macOS, Linux, and Windows, so you're good to
40 go on any platform.
41</Note>
38> [!NOTE]
39>
40> These commands work the same on macOS, Linux, and Windows, so you're good to
41> go on any platform.
4242
4343### Repository Structure
4444
from line 85
8585
8686Start here if you are new to MCP and contributing to its ecosystem.
8787
88<Note>
89 While we use the specification repository as an example, the key patterns are
90 applicable to other MCP repos as well.
91</Note>
88> [!NOTE]
89>
90> While we use the specification repository as an example, the key patterns are
91> applicable to other MCP repos as well.
9292
9393### Step 1: Set Up Your Environment
9494
9595Set up your local environment so you can test and validate changes before submitting them.
9696
97<Steps>
98 <Step title="Fork the repository">
99 Click the **Fork** button on the [repository page](https://github.com/modelcontextprotocol/modelcontextprotocol) to create your own copy. This gives you a personal workspace where you can make changes without affecting the main project.
100 </Step>
97**1. Fork the repository**
10198
102 <Step title="Clone your fork">
103 ```bash theme={null}
104 git clone https://github.com/YOUR-USERNAME/modelcontextprotocol.git
105 cd modelcontextprotocol
106 ```
99Click the **Fork** button on the [repository page](https://github.com/modelcontextprotocol/modelcontextprotocol) to create your own copy. This gives you a personal workspace where you can make changes without affecting the main project.
107100
108 Replace `YOUR-USERNAME` with your GitHub username.
109 </Step>
101**2. Clone your fork**
110102
111 <Step title="Install dependencies">
112 ```bash theme={null}
113 npm install
114 ```
103```bash theme={null}
104git clone https://github.com/YOUR-USERNAME/modelcontextprotocol.git
105cd modelcontextprotocol
106```
115107
116 This installs the tools needed for schema generation, documentation building, and validation.
117 </Step>
108Replace `YOUR-USERNAME` with your GitHub username.
118109
119 <Step title="Verify everything works">
120 ```bash theme={null}
121 npm run check
122 ```
110**3. Install dependencies**
123111
124 This runs TypeScript compilation, schema validation, example validation, documentation link checks, and formatting checks. If everything passes, your environment is good and you're ready to contribute.
125 </Step>
126</Steps>
112```bash theme={null}
113npm install
114```
127115
116This installs the tools needed for schema generation, documentation building, and validation.
117
118**4. Verify everything works**
119
120```bash theme={null}
121npm run check
122```
123
124This runs TypeScript compilation, schema validation, example validation, documentation link checks, and formatting checks. If everything passes, your environment is good and you're ready to contribute.
125
128126If `npm run check` fails, see [Troubleshooting](#troubleshooting) below.
129127
130128### Step 2: Find Something to Work On
from line 142
144142
145143Create your changes in a dedicated branch.
146144
147<Steps>
148 <Step title="Create a branch">
149 ```bash theme={null}
150 git checkout -b fix/your-description
151 ```
145**1. Create a branch**
152146
153 Use a descriptive branch name that reflects your change, like `fix/typo-in-tools-doc` or `feat/add-example-for-resources`.
154 </Step>
147```bash theme={null}
148git checkout -b fix/your-description
149```
155150
156 <Step title="Make your changes">
157 Edit the relevant files in your local copy. If you're editing schema files, remember to run `npm run generate:schema` to regenerate the JSON schema and documentation.
158 </Step>
151Use a descriptive branch name that reflects your change, like `fix/typo-in-tools-doc` or `feat/add-example-for-resources`.
159152
160 <Step title="Run checks">
161 ```bash theme={null}
162 npm run check
163 ```
153**2. Make your changes**
164154
165 Fix any issues before committing. If you have formatting errors, `npm run format` can auto-fix most of them.
166 </Step>
155Edit the relevant files in your local copy. If you're editing schema files, remember to run `npm run generate:schema` to regenerate the JSON schema and documentation.
167156
168 <Step title="Commit with a clear message">
169 ```bash theme={null}
170 git commit -m "Fix typo in tools documentation"
171 ```
157**3. Run checks**
172158
173 Write a concise message that describes what you changed and why. Reference issue numbers if applicable (e.g., `Fix typo in tools documentation (#123)`).
174 </Step>
175</Steps>
159```bash theme={null}
160npm run check
161```
176162
163Fix any issues before committing. If you have formatting errors, `npm run format` can auto-fix most of them.
164
165**4. Commit with a clear message**
166
167```bash theme={null}
168git commit -m "Fix typo in tools documentation"
169```
170
171Write a concise message that describes what you changed and why. Reference issue numbers if applicable (e.g., `Fix typo in tools documentation (#123)`).
172
177173### Step 4: Submit a Pull Request
178174
179175When you're ready, push your branch and open a pull request.
180176
181<Steps>
182 <Step title="Push your branch">
183 ```bash theme={null}
184 git push origin fix/your-description
185 ```
186 </Step>
177**1. Push your branch**
187178
188 <Step title="Open a PR on GitHub">
189 You can use the [GitHub CLI](https://cli.github.com/) to make this process easier:
179```bash theme={null}
180git push origin fix/your-description
181```
190182
191 ```bash theme={null}
192 gh pr create --fill
193 ```
183**2. Open a PR on GitHub**
194184
195 Alternatively, navigate to your fork on GitHub and click **Compare & pull request**.
196 </Step>
185You can use the [GitHub CLI](https://cli.github.com/) to make this process easier:
197186
198 <Step title="Fill in the PR template">
199 Provide a clear description of your changes and link any related issues.
200 </Step>
187```bash theme={null}
188gh pr create --fill
189```
201190
202 <Step title="Wait for review">
203 Maintainers typically respond within 1-5 business days.
204 </Step>
205</Steps>
191Alternatively, navigate to your fork on GitHub and click **Compare & pull request**.
206192
207<Tip>
208 That's it, **congratulations on your first contribution**! Every improvement,
209 no matter how small, helps make MCP better for everyone.
210</Tip>
193**3. Fill in the PR template**
211194
195Provide a clear description of your changes and link any related issues.
196
197**4. Wait for review**
198
199Maintainers typically respond within 1-5 business days.
200
201> [!TIP]
202>
203> That's it, **congratulations on your first contribution**! Every improvement,
204> no matter how small, helps make MCP better for everyone.
205
212206### What Makes a Good Contribution
213207
214208Help us review your contribution quickly by following these patterns:
from line 219
225219
226220Different contributions follow different processes depending on their scope.
227221
228<Tip>
229 Not sure which category your change falls into? Ask in the [MCP Contributor
230 Discord](/community/communication#discord) before starting any significant
231 work.
232</Tip>
222> [!TIP]
223>
224> Not sure which category your change falls into? Ask in the [MCP Contributor
225> Discord](/community/communication#discord) before starting any significant
226> work.
233227
234228### Small Changes (Direct PR)
235229
from line 273
279273
280274To modify the schema:
281275
282<Steps>
283 <Step title="Edit the TypeScript schema">
284 Make your changes in `schema/draft/schema.ts`.
285 </Step>
276**1. Edit the TypeScript schema**
286277
287 <Step title="Add examples (optional)">
288 Add JSON examples in `schema/draft/examples/[TypeName]/` (e.g., `Tool/my-example.json`). Reference them in the schema using `@example` + `@includeCode` JSDoc tags.
289 </Step>
278Make your changes in `schema/draft/schema.ts`.
290279
291 <Step title="Generate JSON schema and docs">
292 ```bash theme={null}
293 npm run generate:schema
294 ```
295 </Step>
280**2. Add examples (optional)**
296281
297 <Step title="Validate your changes">
298 ```bash theme={null}
299 npm run check
300 ```
301 </Step>
302</Steps>
282Add JSON examples in `schema/draft/examples/[TypeName]/` (e.g., `Tool/my-example.json`). Reference them in the schema using `@example` + `@includeCode` JSDoc tags.
303283
284**3. Generate JSON schema and docs**
285
286```bash theme={null}
287npm run generate:schema
288```
289
290**4. Validate your changes**
291
292```bash theme={null}
293npm run check
294```
295
304296### Documentation Changes
305297
306298Docs are written in [MDX format](https://mdxjs.com/) (Markdown with JSX components) and powered by
from line 303
311303
312304Here is how you can contribute to our documentation:
313305
314<Steps>
315 <Step title="Start the local docs server">
316 ```bash theme={null}
317 npm run serve:docs
318 ```
306**1. Start the local docs server**
319307
320 This launches a live preview at `http://localhost:3000` with hot reloading.
321 </Step>
308```bash theme={null}
309npm run serve:docs
310```
322311
323 <Step title="Make your changes">
324 Edit the relevant `.mdx` files. You can use [Mintlify components](https://www.mintlify.com/docs/components) like `<Note>`, `<Tip>`, `<Steps>`, and `<Card>` for richer formatting.
325 </Step>
312This launches a live preview at `http://localhost:3000` with hot reloading.
326313
327 <Step title="Check for issues">
328 ```bash theme={null}
329 npm run check:docs
330 ```
314**2. Make your changes**
331315
332 This validates formatting, broken links, and other common issues.
333 </Step>
334</Steps>
316Edit the relevant `.mdx` files. You can use [Mintlify components](https://www.mintlify.com/docs/components) like `<Note>`, `<Tip>`, `<Steps>`, and `<Card>` for richer formatting.
335317
318**3. Check for issues**
319
320```bash theme={null}
321npm run check:docs
322```
323
324This validates formatting, broken links, and other common issues.
325
336326### Major Protocol Changes
337327
338328For significant changes, follow the [SEP process](/community/sep-guidelines). Prior to spending a
339329lot of time on a spec proposal, make sure to follow these best practices.
340330
341<Steps>
342 <Step title="Validate your idea first">
343 Discuss in an [Interest Group](/community/working-interest-groups) or on
344 [Discord](https://discord.gg/6CSzBmMkjX).
345 </Step>
331**1. Validate your idea first**
346332
347 <Step title="Build a prototype">
348 Demonstrate practical application of your idea.
349 </Step>
333Discuss in an [Interest Group](/community/working-interest-groups) or on
334[Discord](https://discord.gg/6CSzBmMkjX).
350335
351 <Step title="Find a sponsor">
352 A maintainer from the [maintainer
353 list](https://github.com/modelcontextprotocol/modelcontextprotocol/blob/main/MAINTAINERS.md)
354 who will champion your proposal. See [Finding a Sponsor for
355 SEPs](#finding-a-sponsor-for-seps) below.
356 </Step>
336**2. Build a prototype**
357337
358 <Step title="Write the SEP">
359 Follow the [SEP Guidelines](/community/sep-guidelines).
360 </Step>
361</Steps>
338Demonstrate practical application of your idea.
362339
340**3. Find a sponsor**
341
342A maintainer from the [maintainer
343list](https://github.com/modelcontextprotocol/modelcontextprotocol/blob/main/MAINTAINERS.md)
344who will champion your proposal. See [Finding a Sponsor for
345SEPs](#finding-a-sponsor-for-seps) below.
346
347**4. Write the SEP**
348
349Follow the [SEP Guidelines](/community/sep-guidelines).
350
363351### Finding a Sponsor for SEPs
364352
365353A **sponsor** is a Core Maintainer or Maintainer who champions your SEP through the review
from line 354
366354process. They provide feedback, help refine your proposal, and present it at Core Maintainer
367355meetings.
368356
369<Warning>
370 Every SEP needs a sponsor to move forward. SEPs that don't find a sponsor
371 within 6 months are marked as **dormant**. Dormant SEPs aren't rejected
372 outright - they can be revived later if a sponsor is found or the proposal is
373 re-assessed to be needed.
374</Warning>
357> [!WARNING]
358>
359> Every SEP needs a sponsor to move forward. SEPs that don't find a sponsor
360> within 6 months are marked as **dormant**. Dormant SEPs aren't rejected
361> outright - they can be revived later if a sponsor is found or the proposal is
362> re-assessed to be needed.
375363
376364To find a sponsor:
377365
378<Steps>
379 <Step title="Find relevant maintainers">
380 Look at the [maintainer
381 list](https://github.com/modelcontextprotocol/modelcontextprotocol/blob/main/MAINTAINERS.md)
382 to find maintainers working in your area.
383 </Step>
366**1. Find relevant maintainers**
384367
385 <Step title="Tag maintainers in your PR">
386 Tag 1-2 relevant maintainers (don't spam everyone).
387 </Step>
368Look at the [maintainer
369list](https://github.com/modelcontextprotocol/modelcontextprotocol/blob/main/MAINTAINERS.md)
370to find maintainers working in your area.
388371
389 <Step title="Share in Discord">
390 Post your PR in the relevant Discord channel to increase visibility.
391 </Step>
372**2. Tag maintainers in your PR**
392373
393 <Step title="Follow up if needed">
394 If no response after 2 weeks, ask in `#general` or reach out to a Core
395 Maintainer.
396 </Step>
397</Steps>
374Tag 1-2 relevant maintainers (don't spam everyone).
398375
376**3. Share in Discord**
377
378Post your PR in the relevant Discord channel to increase visibility.
379
380**4. Follow up if needed**
381
382If no response after 2 weeks, ask in `#general` or reach out to a Core
383Maintainer.
384
399385Maintainers review open proposals regularly, but response time varies based on complexity and
400386availability.
401387
from line 390
404390MCP maintains official SDKs in multiple languages. Contributions are welcome - whether you're
405391fixing bugs, improving performance, adding features, or enhancing documentation.
406392
407<Note>
408 Each SDK has its own repository, maintainers, and contribution guidelines.
409 Some SDKs are maintained in collaboration with larger partner organizations,
410 such as Google, Microsoft, JetBrains, and others, so processes may vary
411 slightly between repositories.
412</Note>
393> [!NOTE]
394>
395> Each SDK has its own repository, maintainers, and contribution guidelines.
396> Some SDKs are maintained in collaboration with larger partner organizations,
397> such as Google, Microsoft, JetBrains, and others, so processes may vary
398> slightly between repositories.
413399
414400### Before Contributing to an SDK
415401
416402Before diving into code, follow these steps.
417403
418<Steps>
419 <Step title="Open an issue first">
420 Before starting significant work, open an issue to discuss your approach.
421 This helps avoid duplicate effort, ensures your contribution aligns with the
422 SDK's direction, and gives maintainers a chance to provide early feedback.
423 </Step>
404**1. Open an issue first**
424405
425 <Step title="Join the SDK channel">
426 Find the relevant channel in [Discord](https://discord.gg/6CSzBmMkjX) (e.g.,
427 `#typescript-sdk-dev`, `#python-sdk-dev`).
428 </Step>
406Before starting significant work, open an issue to discuss your approach.
407This helps avoid duplicate effort, ensures your contribution aligns with the
408SDK's direction, and gives maintainers a chance to provide early feedback.
429409
430 <Step title="Read the SDK's CONTRIBUTING.md">
431 Each repository has its own `CONTRIBUTING.md` with specific instructions for
432 setting up your development environment, coding standards, commit message
433 conventions, and PR requirements.
434 </Step>
410**2. Join the SDK channel**
435411
436 <Step title="Write tests">
437 All contributions should include appropriate test coverage. Bug fixes should
438 include a test that reproduces the issue, and new features should have tests
439 covering the expected behavior. This helps maintain SDK reliability and
440 prevents regressions.
441 </Step>
442</Steps>
412Find the relevant channel in [Discord](https://discord.gg/6CSzBmMkjX) (e.g.,
413`#typescript-sdk-dev`, `#python-sdk-dev`).
443414
415**3. Read the SDK's CONTRIBUTING.md**
416
417Each repository has its own `CONTRIBUTING.md` with specific instructions for
418setting up your development environment, coding standards, commit message
419conventions, and PR requirements.
420
421**4. Write tests**
422
423All contributions should include appropriate test coverage. Bug fixes should
424include a test that reproduces the issue, and new features should have tests
425covering the expected behavior. This helps maintain SDK reliability and
426prevents regressions.
427
444428### SDK Repositories
445429
446<CardGroup cols={2}>
447 <Card title="TypeScript SDK" icon="square-js" href="https://github.com/modelcontextprotocol/typescript-sdk" />
430> **[TypeScript SDK](https://github.com/modelcontextprotocol/typescript-sdk)**
448431
449 <Card title="Python SDK" icon="python" href="https://github.com/modelcontextprotocol/python-sdk" />
432> **[Python SDK](https://github.com/modelcontextprotocol/python-sdk)**
450433
451 <Card title="Go SDK" icon="golang" href="https://github.com/modelcontextprotocol/go-sdk" />
434> **[Go SDK](https://github.com/modelcontextprotocol/go-sdk)**
452435
453 <Card title="Kotlin SDK" icon="square-k" href="https://github.com/modelcontextprotocol/kotlin-sdk" />
436> **[Kotlin SDK](https://github.com/modelcontextprotocol/kotlin-sdk)**
454437
455 <Card title="Java SDK" icon="java" href="https://github.com/modelcontextprotocol/java-sdk" />
438> **[Java SDK](https://github.com/modelcontextprotocol/java-sdk)**
456439
457 <Card title="C# SDK" icon="square-c" href="https://github.com/modelcontextprotocol/csharp-sdk" />
440> **[C# SDK](https://github.com/modelcontextprotocol/csharp-sdk)**
458441
459 <Card title="Swift SDK" icon="swift" href="https://github.com/modelcontextprotocol/swift-sdk" />
442> **[Swift SDK](https://github.com/modelcontextprotocol/swift-sdk)**
460443
461 <Card title="Rust SDK" icon="rust" href="https://github.com/modelcontextprotocol/rust-sdk" />
444> **[Rust SDK](https://github.com/modelcontextprotocol/rust-sdk)**
462445
463 <Card title="Ruby SDK" icon="gem" href="https://github.com/modelcontextprotocol/ruby-sdk" />
446> **[Ruby SDK](https://github.com/modelcontextprotocol/ruby-sdk)**
464447
465 <Card title="PHP SDK" icon="php" href="https://github.com/modelcontextprotocol/php-sdk" />
466</CardGroup>
448> **[PHP SDK](https://github.com/modelcontextprotocol/php-sdk)**
467449
468450## Getting Help
469451
community/contributor-ladder Changed · +12 / -12 lines
from line 24
2424| [**Lead Maintainer**](#lead-maintainer) | Ultimate project authority (founders) | All Core Maintainer privileges, veto authority, appoints Core Maintainers | Reserved for project founders; succession only |
2525| [**Community Moderator**](#community-moderator) | CoC enforcement and community health | Moderation rights on community platforms, incident handling | Parallel track: Member status + appointment |
2626
27<Note>
28 Timelines are minimums, not guarantees. They protect the project from rapid
29 privilege escalation and ensure a high bar of demonstrated commitment. Actual
30 advancement is discretionary and may take longer. Exceptions require explicit
31 Core Maintainer approval with documented rationale.
32</Note>
27> [!NOTE]
28>
29> Timelines are minimums, not guarantees. They protect the project from rapid
30> privilege escalation and ensure a high bar of demonstrated commitment. Actual
31> advancement is discretionary and may take longer. Exceptions require explicit
32> Core Maintainer approval with documented rationale.
3333
3434***
3535
from line 134
134134
135135Core Maintainers hold final decision-making authority for MCP's technical direction. This is the highest level of trust in the community.
136136
137<Note>
138 The Core Maintainer role is intentionally limited. This ensures a coherent
139 technical vision while the project scales. Bandwidth concerns are addressed
140 through delegation to Maintainers, Working Group Leads, and Interest Group
141 Facilitators, not by expanding Core Maintainer numbers.
142</Note>
137> [!NOTE]
138>
139> The Core Maintainer role is intentionally limited. This ensures a coherent
140> technical vision while the project scales. Bandwidth concerns are addressed
141> through delegation to Maintainers, Working Group Leads, and Interest Group
142> Facilitators, not by expanding Core Maintainer numbers.
143143
144144**Requirements:**
145145
community/interest-groups/agents New page · 108 lines, new page
# Agents Charter ## Group Type ## Mission Statement ## Scope ### In Scope ### Out of Scope ### Related Groups ## Leadership ## Membership ## Operations ## Resources ## Discussion Topics ### Planned Outputs ## Changelog
A whole new page. There's nothing to diff it against, so here is what it says.
# Agents Charter > Charter for the MCP Agents Interest Group. ## Group Type **Interest Group** ## Mission Statement The Agents Interest Group exists to make interactions with agent-backed systems interoperable over MCP. Today, these systems are typically exposed as ordinary tools or through framework-specific integrations, leaving durable execution, capability discovery, delegation, and multi-turn interaction to ad hoc conventions. The IG evaluates which remaining gaps require protocol support. Based on production use cases and prototypes, the group will recommend how to evolve the protocol, create additional extensions or document how existing MCP primitives should be used consistently. ## Scope ### In Scope * **Agents Extension Evaluation**: Evaluate agent-backed MCP use cases, including agent-as-tool, remote-agent, and supervisor/sub-agent patterns, to determine whether an Agents Extension is needed to complement Tasks. * **Prototypes and Proposals**: Develop narrowly scoped proofs of concept where they help determine whether to introduce an Agents Extension, evolve Tasks, or rely on existing MCP primitives. * **Cross-Cutting Concerns**: Coordinate with relevant groups when evaluated agent use cases raise tasks, transport, events, skills, authorization, metadata, or external-interoperability questions. * **Documentation**: Document agent-backed MCP use cases, findings, and any recommendations produced by the group. ### Out of Scope * Building or standardizing general-purpose agent frameworks and runtimes, including internal choices about planning, memory, model selection, and orchestration. The group focuses on behavior at MCP interoperability boundaries rather than host or server implementation internals. * Transport wire formats and session mechanics, which are owned by the Transports WG. * Tasks, general event delivery, and callback mechanisms, which are owned by the Tasks and Events WG. The charter does not predetermine where inference or agent loops run, or whether a particular capability belongs in Tasks, an Agents Extension, or existing MCP primitives. The IG evaluates those questions through its discussions. ### Related Groups * **[Transports WG](/community/working-groups/transports)** - Task polling, multi-round-trip requests, stateless operation, and request association depend on transport and message-flow semantics. * **[Tasks and Events WG](/community/working-groups/tasks-events)** - Tasks and proactive task-status and completion notifications are owned by that group. The IG provides use cases and requirements as input. ## Leadership | Role | Name | Organization | GitHub | Term | | - | - | - | - | - | | Facilitator | Luca Chang | Amazon Web Services | [@LucaButBoring](https://github.com/LucaButBoring) | Initial | ## Membership | Name | Organization | GitHub | Discord | Level | | - | - | - | - | - | | Luca Chang | Amazon Web Services | [@LucaButBoring](https://github.com/LucaButBoring) | | Facilitator | ## Operations | Meeting | Frequency | Duration | Purpose | | - | - | - | - | | Working Session | Weekly | 30 minutes | Technical discussion, research, and proposal review | Meetings are published at [meet.modelcontextprotocol.io](https://meet.modelcontextprotocol.io). Discord: `#agents-wg` ## Resources * Group repository: [modelcontextprotocol/agents-wg](https://github.com/modelcontextprotocol/agents-wg) ## Discussion Topics | Item | Status | Target Date | Champion | | - | - | - | - | | Agents Extension evaluation and recommendation | In Progress | | TBD | | Two-level agent definition proof of concept | In Progress | | [@LucaButBoring](https://github.com/LucaButBoring), [@madhaviai](https://github.com/madhaviai) | ### Planned Outputs * The group publishes an evaluation of whether an Agents Extension is needed to complement Tasks, supported by documented use cases and relevant prototype evidence. * Based on that evaluation, the group recommends evolving Tasks, recommends an Agents Extension, or documents how existing MCP primitives are sufficient. * Proposed agent protocol surface is evaluated through prototypes and implementation feedback before being advanced. ## Changelog | Date | Change | | - | - | | 2026-08-04 | Initial charter (as the Agents WG) | | 2026-10-16 | Converted to an Interest Group. Tasks scope moved to the Tasks and Events WG. |
community/interest-groups/enterprise-managed-authorization Changed · +7 / -7 lines
from line 2
22
33> Charter for the MCP Enterprise-Managed Authorization Interest Group.
44
5<Note>
6 As of 2026-08-17 this Interest Group is folded into the [Authorization
7 IG](/community/interest-groups/auth). EMA interoperability and deployment
8 progress is presented as agenda slots on the Auth IG call, discussion
9 continues in `#auth-ig` threads, and `#enterprise-managed-auth-ig` is
10 archived. This page is kept for reference.
11</Note>
5> [!NOTE]
6>
7> As of 2026-08-17 this Interest Group is folded into the [Authorization
8> IG](/community/interest-groups/auth). EMA interoperability and deployment
9> progress is presented as agenda slots on the Auth IG call, discussion
10> continues in `#auth-ig` threads, and `#enterprise-managed-auth-ig` is
11> archived. This page is kept for reference.
1212
1313## Group Type
1414
community/sdk-tiers Changed · +10 / -10 lines
from line 4
44
55The MCP SDK Tiering System establishes clear expectations for feature completeness, protocol support, and maintenance commitments across official and community-driven SDKs. This helps developers choose the right SDK for their needs and provides SDK maintainers with a clear path to improving adoption expectations.
66
7<Note>
8 **Key dates:**
9
10 * **January 23, 2026**: Conformance tests available
11 * **February 23, 2026**: Official SDK tiering published
12
13 Between January 23 and February 23, SDK maintainers can work with the
14 Conformance Testing working group to adopt the tests and set up GitHub issue
15 tracking with the standardized labels defined below.
16</Note>
7> [!NOTE]
8>
9> **Key dates:**
10>
11> * **January 23, 2026**: Conformance tests available
12> * **February 23, 2026**: Official SDK tiering published
13>
14> Between January 23 and February 23, SDK maintainers can work with the
15> Conformance Testing working group to adopt the tests and set up GitHub issue
16> tracking with the standardized labels defined below.
1717
1818## Overview
1919
community/sep-guidelines Changed · +8 / -8 lines
from line 71
7171
7272**Prior discussion is required.** SEP authors must discuss the idea with the relevant [working or interest group](/community/working-interest-groups) in [Discord](/community/communication#discord) before opening the SEP pull request, and the PR description must link to that discussion (a Discord thread, WG/IG meeting notes, or a GitHub Discussion). A SEP without a linked prior discussion is not accepted. If no relevant group exists, start a conversation in [GitHub Discussions](https://github.com/modelcontextprotocol/modelcontextprotocol/discussions) or the `#general` channel in [Discord](/community/communication#discord) instead. If there is enough interest, it may be worth [creating a new IG or WG](/community/working-interest-groups#creating-an-interest-group). The effort involved in finding sponsors and facilitators is a good signal of whether the idea has sufficient traction.
7373
74<Note>
75 To improve your chances of a SEP being accepted, check alignment with [Core
76 Maintainer](/community/governance#roles) priorities and [design
77 principles](/community/design-principles). Priorities are generally reflected
78 in the [project roadmap](/development/roadmap). Proposals outside current
79 priorities or that conflict with design principles are more likely to face
80 delays or additional friction in the review process.
81</Note>
74> [!NOTE]
75>
76> To improve your chances of a SEP being accepted, check alignment with [Core
77> Maintainer](/community/governance#roles) priorities and [design
78> principles](/community/design-principles). Priorities are generally reflected
79> in the [project roadmap](/development/roadmap). Proposals outside current
80> priorities or that conflict with design principles are more likely to face
81> delays or additional friction in the review process.
8282
83831. **Discuss with the relevant group**: Bring your idea to the matching [working or interest group](/community/working-interest-groups) and refine it there. Keep a link to the discussion (a Discord thread, meeting notes, or a GitHub Discussion) so you can include it in the PR description.
8484
community/working-groups/agents Page removed · 143 lines, page removed
# Agents Charter ## Group Type ## Mission Statement ## Scope ### In Scope ### Out of Scope ### Related Groups ## Leadership ## Authority & Decision Rights ## Membership ## Operations ## Resources ## Deliverables & Success Metrics ### Active Work Items ### Success Criteria ## Changelog
The page is gone upstream. What it last said is kept here.
community/working-groups/tasks-events New page · 112 lines, new page
# Tasks and Events Charter ## Group Type ## Mission Statement ## Scope ### In Scope ### Out of Scope ### Related Groups ## Leadership ## Authority & Decision Rights ## Membership ## Operations ## Resources ## Deliverables & Success Metrics ### Active Work Items ### Success Criteria ## Changelog
A whole new page. There's nothing to diff it against, so here is what it says.
# Tasks and Events Charter > Charter for the MCP Tasks and Events Working Group. ## Group Type **Working Group** ## Mission Statement The Tasks and Events Working Group exists to continue the development of Tasks, MCP's existing foundation for durable asynchronous execution, and to define how servers proactively notify clients of state changes. Today, clients learn about server-side updates by polling or holding an SSE connection open. This WG will specify a standardized callback mechanism—webhooks or similar—that lets servers push notifications when new data is available, with defined ordering guarantees that hold across all transports. ## Scope ### In Scope * **Tasks**: Stabilization of the [`io.modelcontextprotocol/tasks`](https://github.com/modelcontextprotocol/modelcontextprotocol/pull/2663) extension and its promotion into the core MCP protocol, including collecting implementation feedback, resolving ambiguities, and evaluating proposed changes to its lifecycle. * **Specification Work**: SEPs defining the trigger/callback mechanism, subscription lifecycle, delivery semantics, and event ordering guarantees. * **Reference Implementations**: SDK components demonstrating server-initiated notifications and client-side callback handling. * **Cross-Cutting Concerns**: Coordination with the Transports WG on transport-specific delivery behavior, and with the Agents IG on agent-backed use cases that depend on Tasks or event triggers. * **Documentation**: The Tasks specification, and specification sections covering event-driven patterns and migration guidance from polling-based approaches. ### Out of Scope * Changes to the transport wire format or session model (owned by the Transports WG). * General-purpose pub/sub infrastructure beyond what the MCP protocol requires. * Modifications to existing notification primitives (`notifications/resources/updated`, `notifications/tools/list_changed`, etc.) that do not relate to proactive server-initiated delivery. * Evaluation of agent-backed MCP use cases and of a possible Agents Extension, which belongs to the Agents IG. ### Related Groups * **[Transports WG](/community/working-groups/transports)** — delivery and ordering guarantees depend on transport capabilities; callback semantics must be coherent across stdio, Streamable HTTP, and future transports. * **[Agents IG](/community/interest-groups/agents)** — agent-as-tool, remote-agent, and supervisor/sub-agent patterns depend on Tasks and on task completion notifications. The IG provides use cases and requirements, and this WG owns the protocol surface. ## Leadership | Role | Name | Organization | GitHub | Term | | - | - | - | - | - | | Lead | Clare Liguori | Amazon Web Services | [@clareliguori](https://github.com/clareliguori) | Initial | | Lead | Peter Alexander | Anthropic | [@pja-ant](https://github.com/pja-ant) | Initial | | Lead | Luca Chang | Amazon Web Services | [@LucaButBoring](https://github.com/LucaButBoring) | Initial | | Lead | Caitie McCaffrey | Microsoft | [@CaitieM20](https://github.com/CaitieM20) | Initial | ## Authority & Decision Rights | Decision Type | Authority Level | | - | - | | Meeting logistics & scheduling | WG Leads (autonomous) | | Proposal prioritization within WG | WG Leads (autonomous) | | SEP triage & closure (in scope) | WG Leads (autonomous, with documented rationale) | | Technical design within scope | WG consensus | | Spec changes (additive) | WG consensus → Core Maintainer approval | | Spec changes (breaking/fundamental) | WG consensus → Core Maintainer approval + wider review | | Scope expansion | Core Maintainer approval required | | WG Member approval | WG Member sponsors | ## Membership | Name | Organization | GitHub | Discord | Level | | - | - | - | - | - | | Clare Liguori | Amazon Web Services | [@clareliguori](https://github.com/clareliguori) | | Lead | | Peter Alexander | Anthropic | [@pja-ant](https://github.com/pja-ant) | | Lead | | Luca Chang | Amazon Web Services | [@LucaButBoring](https://github.com/LucaButBoring) | | Lead | | Caitie McCaffrey | Microsoft | [@CaitieM20](https://github.com/CaitieM20) | | Lead | ## Operations | Meeting | Frequency | Duration | Purpose | | - | - | - | - | | Working Session | Weekly | 30 min | Technical discussion, proposal review | Meetings are published at [meet.modelcontextprotocol.io](https://meet.modelcontextprotocol.io). ## Resources * Incubation repository: [modelcontextprotocol/experimental-ext-triggers-events](https://github.com/modelcontextprotocol/experimental-ext-triggers-events) * Tasks extension repository: [modelcontextprotocol/ext-tasks](https://github.com/modelcontextprotocol/ext-tasks) * Tasks specification: [SEP-2663: Tasks Extension](/seps/2663-tasks-extension) ## Deliverables & Success Metrics ### Active Work Items | Item | Status | Target Date | Champion | | - | - | - | - | | Tasks stabilization and core protocol promotion | In Progress | | [@LucaButBoring](https://github.com/LucaButBoring) | | SEP: Events in MCP v1 RFC | Ideating | End April | TBD | | Reference implementation in Tier-1 SDKs | — | End April | TBD | ### Success Criteria * Tasks is fully stabilized based on implementation feedback and promoted from an extension into the core MCP protocol. * Lifecycle ambiguities and conformance gaps that block Tasks stabilization are resolved or explicitly dispositioned. * An accepted SEP defining the trigger/callback mechanism and its subscription lifecycle. * Reference implementations in at least two Tier-1 SDKs. * Conformance test coverage for the new primitives. ## Changelog | Date | Change | | - | - | | 2026-03-24 | Initial charter (as the Triggers and Events WG) | | 2026-10-16 | Renamed to Tasks and Events WG. Tasks scope and its leads moved here from the Agents WG. |
community/working-groups/transports Changed · +6 / -6 lines
from line 67
6767* **[SDK WG](/community/working-groups/sdk)**: Official SDKs implement transport
6868 changes; the groups coordinate on feasibility, reference implementations, and
6969 rollout sequencing.
70* **[Agents WG](/community/working-groups/agents)**: The Agents WG owns Tasks and
71 durable execution; the groups coordinate where these depend on multi-round-trip
72 requests, request association, stateless operation, or binding behavior.
73* **[Triggers and Events WG](/community/working-groups/triggers-events)**: Event
74 callbacks, subscriptions, and application-level delivery semantics are owned by that
75 group; Transports owns binding-specific carriage and guarantees.
70* **[Tasks and Events WG](/community/working-groups/tasks-events)**: The Tasks and
71 Events WG owns Tasks and durable execution; the groups coordinate where these
72 depend on multi-round-trip requests, request association, stateless operation, or
73 binding behavior. Event callbacks, subscriptions, and application-level delivery
74 semantics are owned by that group; Transports owns binding-specific carriage and
75 guarantees.
7676* **[Auth IG](/community/interest-groups/auth)** and
7777 **[Security IG](/community/interest-groups/security)**: The groups coordinate on
7878 credential carriage and transport wire security while authorization mechanics and
community/working-groups/triggers-events Page removed · 82 lines, page removed
# Triggers and Events Charter ## Group Type ## Mission Statement ## Scope ### In Scope ### Out of Scope ### Related Groups ## Leadership ## Authority & Decision Rights ## Operations ## Resources ## Deliverables & Success Metrics ### Active Work Items ### Success Criteria ## Changelog
The page is gone upstream. What it last said is kept here.
community/working-interest-groups Changed · +20 / -22 lines
from line 142
142142
143143**WG Consensus** is achieved through the following progression. Each step is attempted before moving to the next.
144144
145<Steps>
146 <Step title="Lazy Consensus (default)">
147 * Proposals announced with clear deadline (5 days minimum for minor items, 10 days for significant items)
148 * Silence is consent
149 * Any WG Member may block with documented objection
150 * Blocks must propose alternatives or clear criteria for resolution
151 * If no blocks are raised by the deadline, the proposal is accepted
152 </Step>
145**1. Lazy Consensus (default)**
153146
154 <Step title="Formal Vote (when lazy consensus is blocked)">
155 A formal vote is triggered when a WG Member blocks during the lazy consensus period, or when a Lead or three or more WG Members request one.
147* Proposals announced with clear deadline (5 days minimum for minor items, 10 days for significant items)
148* Silence is consent
149* Any WG Member may block with documented objection
150* Blocks must propose alternatives or clear criteria for resolution
151* If no blocks are raised by the deadline, the proposal is accepted
156152
157 * Quorum: 50% of active WG Members
158 * Passage: simple majority for routine matters; 2/3 majority for scope changes
159 * Core Maintainer feedback is advisory unless explicitly stated as binding
160 * All votes documented with rationale
161 </Step>
153**2. Formal Vote (when lazy consensus is blocked)**
162154
163 <Step title="Escalation (when voting does not resolve)">
164 If a vote fails to resolve the matter (no quorum, does not pass, or the result is contested), the Lead escalates to Core Maintainers following the escalation path below.
165 </Step>
166</Steps>
155A formal vote is triggered when a WG Member blocks during the lazy consensus period, or when a Lead or three or more WG Members request one.
156
157* Quorum: 50% of active WG Members
158* Passage: simple majority for routine matters; 2/3 majority for scope changes
159* Core Maintainer feedback is advisory unless explicitly stated as binding
160* All votes documented with rationale
161
162**3. Escalation (when voting does not resolve)**
163
164If a vote fails to resolve the matter (no quorum, does not pass, or the result is contested), the Lead escalates to Core Maintainers following the escalation path below.
167165
168166### Escalation Path
169167
development/roadmap Changed · +8 / -6 lines
from line 2
22
33> The Model Context Protocol roadmap, covering the priority areas Core Maintainers and Working Groups are driving for the next specification release and how they shape SEP review.
44
5<Info>Last updated: **2026-08-22**</Info>
5> [!NOTE]
6>
7> Last updated: **2026-08-22**
68
79This page outlines the Core Maintainers' vision for the protocol over the coming six to twelve months, highlighting objectives targeted for the next specification update and beyond. It describes our primary strategic goals and the expected deliverables from [Working and Interest Groups](/community/working-interest-groups).
810
9<Note>
10 This roadmap reflects current thinking rather than firm commitments. Priorities may shift, some items may be delivered differently than described or deferred, and work not listed here may still be included in the release.
11</Note>
11> [!NOTE]
12>
13> This roadmap reflects current thinking rather than firm commitments. Priorities may shift, some items may be delivered differently than described or deferred, and work not listed here may still be included in the release.
1214
1315## SEP Prioritization
1416
from line 30
2830
2931**This roadmap period:**
3032
31* **Server-initiated events**: [Triggers & Events WG](/community/working-groups/triggers-events). Channels and subscriptions for push delivery, including webhooks. As we take on asynchronous workloads through Tasks and other events, we need extensions that let servers tell clients when work has finished, without relying purely on expensive client-side polling.
32* **A composition review**: [Agents](/community/working-groups/agents), Transports, and [Triggers & Events](/community/working-groups/triggers-events) WGs. Primitives in the making, such as Tasks and Triggers, need to compose cleanly with each other and fit concrete use cases.
33* **Server-initiated events**: [Tasks & Events WG](/community/working-groups/tasks-events). Channels and subscriptions for push delivery, including webhooks. As we take on asynchronous workloads through Tasks and other events, we need extensions that let servers tell clients when work has finished, without relying purely on expensive client-side polling.
34* **A composition review**: [Tasks & Events](/community/working-groups/tasks-events) and Transports WGs, with the [Agents](/community/interest-groups/agents) IG. Primitives in the making, such as Tasks and Triggers, need to compose cleanly with each other and fit concrete use cases.
3335
3436Beyond these, we expect continued work on Tasks ([SEP-2663](/seps/2663-tasks-extension)) toward eventual inclusion of the extension in the core protocol.
3537
docs/2024-11-05/develop/build-client Changed · +2436 / -2452 lines
## System Requirements ## Setting Up Your Environment ## Setting Up Your API Key ## Creating the Client ### Basic Client Structure ### Server Connection Management ### Query Processing Logic ### Interactive Chat Interface ### Main Entry Point ## Key Components Explained ### 1. Client Initialization ### 2. Server Connection ### 3. Query Processing ### 4. Interactive Interface ### 5. Resource Management ## Common Customization Points ## Running the Client ## How It Works ## Best practices ## Troubleshooting ### Server Path Issues ### Response Timing ### Common Error Messages ## System Requirements ## Setting Up Your Environment ## Setting Up Your API Key ## Creating the Client ### Basic Client Structure ### Server Connection Management ### Query Processing Logic ### Interactive Chat Interface ### Main Entry Point ## Running the Client ## How It Works ## Best practices ## Troubleshooting ### Server Path Issues ### Response Timing ### Common Error Messages ## System Requirements ## Setting Up Your Environment ## How it Works ### MCP Client Configuration ### Chat Implementation ### Build and run ### Advanced Configuration ## System Requirements ## Setting up your environment ## Setting up your API key ## Creating the Client ### Basic Client Structure ### Server connection management ### Query processing logic ### Interactive chat ### Main entry point ## Running the client ## How it works ## Best practices ## Troubleshooting ### Server Path Issues ### Build Issues ### Response Timing ### Common Error Messages ## System Requirements ## Setting up your environment ## Setting up your API key ## Creating the Client ### Basic Client Structure ### Query processing logic ## Key Components Explained ### 1. Client Initialization ### 2. Server Connection ### 3. Query Processing ## Running the Client ## System Requirements ## Setting Up Your Environment ## Setting Up Your API Key ## Creating the Client ### Basic Client Structure ### Server Connection Management ### Query Processing Logic ### Interactive Chat Interface ### Main Entry Point ## Key Components Explained ### 1. Client Initialization ### 2. Server Connection ### 3. Query Processing ### 4. Interactive Interface ### 5. Resource Management ## Running the Client ## How It Works ## Best practices ## Troubleshooting ### Server Path Issues ### Response Timing ### Common Error Messages ## System Requirements ## Setting Up Your Environment ## Setting Up Your API Key ## Creating the Client ### Imports and Client Structure ### Client Initialization ### Server Connection Management ### Converting MCP Tools ### Sending Model Requests ### Query Processing Logic ### Interactive Chat Interface ### Cleanup ### Main Entry Point ### Verify the Complete File ## Running the Client ## How It Works ## Best Practices ## Troubleshooting ### Server Command Issues ### Environment File Issues ### Tool and Response Errors
The two sides of this change are more than 400 edits apart, too far apart to line up, so this is the differ's own diff of it and the words inside a line are not marked.
from line 6
66
77Before you begin, it helps to have gone through our [Build an MCP Server](/docs/2024-11-05/develop/build-server) tutorial so you can understand how clients and servers communicate.
88
9<Tabs>
10 <Tab title="Python">
11 [You can find the complete code for this tutorial here.](https://github.com/modelcontextprotocol/quickstart-resources/tree/main/mcp-client-python)
12
13 ## System Requirements
14
15 Before starting, ensure your system meets these requirements:
16
17 * Mac or Windows computer
18 * Latest Python version installed
19 * Latest version of `uv` installed
20
21 ## Setting Up Your Environment
22
23 First, create a new Python project with `uv`:
24
25 <CodeGroup>
26 ```bash macOS/Linux theme={null}
27 # Create project directory
28 uv init mcp-client
29 cd mcp-client
30
31 # Create virtual environment
32 uv venv
33
34 # Activate virtual environment
35 source .venv/bin/activate
36
37 # Install required packages
38 uv add mcp anthropic python-dotenv
39
40 # Remove boilerplate files
41 rm main.py
42
43 # Create our main file
44 touch client.py
45 ```
46
47 ```powershell Windows theme={null}
48 # Create project directory
49 uv init mcp-client
50 cd mcp-client
51
52 # Create virtual environment
53 uv venv
54
55 # Activate virtual environment
56 .venv\Scripts\activate
57
58 # Install required packages
59 uv add mcp anthropic python-dotenv
60
61 # Remove boilerplate files
62 del main.py
63
64 # Create our main file
65 new-item client.py
66 ```
67 </CodeGroup>
68
69 ## Setting Up Your API Key
70
71 You'll need an Anthropic API key from the [Anthropic Console](https://console.anthropic.com/settings/keys).
72
73 Create a `.env` file to store it:
74
75 ```bash theme={null}
76 echo "ANTHROPIC_API_KEY=your-api-key-goes-here" > .env
77 ```
78
79 Add `.env` to your `.gitignore`:
80
81 ```bash theme={null}
82 echo ".env" >> .gitignore
83 ```
84
85 <Warning>
86 Make sure you keep your `ANTHROPIC_API_KEY` secure!
87 </Warning>
88
89 ## Creating the Client
90
91 ### Basic Client Structure
92
93 First, let's set up our imports and create the basic client class:
94
95 ```python theme={null}
96 import asyncio
97 from typing import Optional
98 from contextlib import AsyncExitStack
99
100 from mcp import ClientSession, StdioServerParameters
101 from mcp.client.stdio import stdio_client
102
103 from anthropic import Anthropic
104 from dotenv import load_dotenv
105
106 load_dotenv() # load environment variables from .env
107
108 class MCPClient:
109 def __init__(self):
110 # Initialize session and client objects
111 self.session: Optional[ClientSession] = None
112 self.exit_stack = AsyncExitStack()
113 self.anthropic = Anthropic()
114 # methods will go here
115 ```
116
117 ### Server Connection Management
118
119 Next, we'll implement the method to connect to an MCP server:
120
121 ```python theme={null}
122 async def connect_to_server(self, server_script_path: str):
123 """Connect to an MCP server
124
125 Args:
126 server_script_path: Path to the server script (.py or .js)
127 """
128 is_python = server_script_path.endswith('.py')
129 is_js = server_script_path.endswith('.js')
130 if not (is_python or is_js):
131 raise ValueError("Server script must be a .py or .js file")
132
133 command = "python" if is_python else "node"
134 server_params = StdioServerParameters(
135 command=command,
136 args=[server_script_path],
137 env=None
138 )
139
140 stdio_transport = await self.exit_stack.enter_async_context(stdio_client(server_params))
141 self.stdio, self.write = stdio_transport
142 self.session = await self.exit_stack.enter_async_context(ClientSession(self.stdio, self.write))
143
144 await self.session.initialize()
145
146 # List available tools
147 response = await self.session.list_tools()
148 tools = response.tools
149 print("\nConnected to server with tools:", [tool.name for tool in tools])
150 ```
151
152 ### Query Processing Logic
153
154 Now let's add the core functionality for processing queries and handling tool calls:
155
156 ```python theme={null}
157 async def process_query(self, query: str) -> str:
158 """Process a query using Claude and available tools"""
159 messages = [
160 {
9**Python**
10
11[You can find the complete code for this tutorial here.](https://github.com/modelcontextprotocol/quickstart-resources/tree/main/mcp-client-python)
12
13## System Requirements
14
15Before starting, ensure your system meets these requirements:
16
17* Mac or Windows computer
18* Latest Python version installed
19* Latest version of `uv` installed
20
21## Setting Up Your Environment
22
23First, create a new Python project with `uv`:
24
25```bash macOS/Linux theme={null}
26# Create project directory
27uv init mcp-client
28cd mcp-client
29
30# Create virtual environment
31uv venv
32
33# Activate virtual environment
34source .venv/bin/activate
35
36# Install required packages
37uv add mcp anthropic python-dotenv
38
39# Remove boilerplate files
40rm main.py
41
42# Create our main file
43touch client.py
44```
45
46```powershell Windows theme={null}
47# Create project directory
48uv init mcp-client
49cd mcp-client
50
51# Create virtual environment
52uv venv
53
54# Activate virtual environment
55.venv\Scripts\activate
56
57# Install required packages
58uv add mcp anthropic python-dotenv
59
60# Remove boilerplate files
61del main.py
62
63# Create our main file
64new-item client.py
65```
66
67## Setting Up Your API Key
68
69You'll need an Anthropic API key from the [Anthropic Console](https://console.anthropic.com/settings/keys).
70
71Create a `.env` file to store it:
72
73```bash theme={null}
74echo "ANTHROPIC_API_KEY=your-api-key-goes-here" > .env
75```
76
77Add `.env` to your `.gitignore`:
78
79```bash theme={null}
80echo ".env" >> .gitignore
81```
82
83> [!WARNING]
84>
85> Make sure you keep your `ANTHROPIC_API_KEY` secure!
86
87## Creating the Client
88
89### Basic Client Structure
90
91First, let's set up our imports and create the basic client class:
92
93```python theme={null}
94import asyncio
95from typing import Optional
96from contextlib import AsyncExitStack
97
98from mcp import ClientSession, StdioServerParameters
99from mcp.client.stdio import stdio_client
100
101from anthropic import Anthropic
102from dotenv import load_dotenv
103
104load_dotenv() # load environment variables from .env
105
106class MCPClient:
107 def __init__(self):
108 # Initialize session and client objects
109 self.session: Optional[ClientSession] = None
110 self.exit_stack = AsyncExitStack()
111 self.anthropic = Anthropic()
112 # methods will go here
113```
114
115### Server Connection Management
116
117Next, we'll implement the method to connect to an MCP server:
118
119```python theme={null}
120async def connect_to_server(self, server_script_path: str):
121 """Connect to an MCP server
122
123 Args:
124 server_script_path: Path to the server script (.py or .js)
125 """
126 is_python = server_script_path.endswith('.py')
127 is_js = server_script_path.endswith('.js')
128 if not (is_python or is_js):
129 raise ValueError("Server script must be a .py or .js file")
130
131 command = "python" if is_python else "node"
132 server_params = StdioServerParameters(
133 command=command,
134 args=[server_script_path],
135 env=None
136 )
137
138 stdio_transport = await self.exit_stack.enter_async_context(stdio_client(server_params))
139 self.stdio, self.write = stdio_transport
140 self.session = await self.exit_stack.enter_async_context(ClientSession(self.stdio, self.write))
141
142 await self.session.initialize()
143
144 # List available tools
145 response = await self.session.list_tools()
146 tools = response.tools
147 print("\nConnected to server with tools:", [tool.name for tool in tools])
148```
149
150### Query Processing Logic
151
152Now let's add the core functionality for processing queries and handling tool calls:
153
154```python theme={null}
155async def process_query(self, query: str) -> str:
156 """Process a query using Claude and available tools"""
157 messages = [
158 {
159 "role": "user",
160 "content": query
161 }
162 ]
163
164 response = await self.session.list_tools()
165 available_tools = [{
166 "name": tool.name,
167 "description": tool.description,
168 "input_schema": tool.inputSchema
169 } for tool in response.tools]
170
171 # Initial Claude API call
172 response = self.anthropic.messages.create(
173 model="claude-sonnet-4-20250514",
174 max_tokens=1000,
175 messages=messages,
176 tools=available_tools
177 )
178
179 # Process response and handle tool calls
180 final_text = []
181
182 assistant_message_content = []
183 for content in response.content:
184 if content.type == 'text':
185 final_text.append(content.text)
186 assistant_message_content.append(content)
187 elif content.type == 'tool_use':
188 tool_name = content.name
189 tool_args = content.input
190
191 # Execute tool call
192 result = await self.session.call_tool(tool_name, tool_args)
193 final_text.append(f"[Calling tool {tool_name} with args {tool_args}]")
194
195 assistant_message_content.append(content)
196 messages.append({
197 "role": "assistant",
198 "content": assistant_message_content
199 })
200 messages.append({
161201 "role": "user",
162 "content": query
163 }
164 ]
165
166 response = await self.session.list_tools()
167 available_tools = [{
168 "name": tool.name,
169 "description": tool.description,
170 "input_schema": tool.inputSchema
171 } for tool in response.tools]
172
173 # Initial Claude API call
174 response = self.anthropic.messages.create(
175 model="claude-sonnet-4-20250514",
176 max_tokens=1000,
177 messages=messages,
178 tools=available_tools
179 )
180
181 # Process response and handle tool calls
182 final_text = []
183
184 assistant_message_content = []
185 for content in response.content:
186 if content.type == 'text':
187 final_text.append(content.text)
188 assistant_message_content.append(content)
189 elif content.type == 'tool_use':
190 tool_name = content.name
191 tool_args = content.input
192
193 # Execute tool call
194 result = await self.session.call_tool(tool_name, tool_args)
195 final_text.append(f"[Calling tool {tool_name} with args {tool_args}]")
196
197 assistant_message_content.append(content)
198 messages.append({
199 "role": "assistant",
200 "content": assistant_message_content
201 })
202 messages.append({
203 "role": "user",
204 "content": [
205 {
206 "type": "tool_result",
207 "tool_use_id": content.id,
208 "content": result.content
209 }
210 ]
211 })
212
213 # Get next response from Claude
214 response = self.anthropic.messages.create(
215 model="claude-sonnet-4-20250514",
216 max_tokens=1000,
217 messages=messages,
218 tools=available_tools
219 )
220
221 final_text.append(response.content[0].text)
222
223 return "\n".join(final_text)
224 ```
225
226 ### Interactive Chat Interface
227
228 Now we'll add the chat loop and cleanup functionality:
229
230 ```python theme={null}
231 async def chat_loop(self):
232 """Run an interactive chat loop"""
233 print("\nMCP Client Started!")
234 print("Type your queries or 'quit' to exit.")
235
236 while True:
237 try:
238 query = input("\nQuery: ").strip()
239
240 if query.lower() == 'quit':
241 break
242
243 response = await self.process_query(query)
244 print("\n" + response)
245
246 except Exception as e:
247 print(f"\nError: {str(e)}")
248
249 async def cleanup(self):
250 """Clean up resources"""
251 await self.exit_stack.aclose()
252 ```
253
254 ### Main Entry Point
255
256 Finally, we'll add the main execution logic:
257
258 ```python theme={null}
259 async def main():
260 if len(sys.argv) < 2:
261 print("Usage: python client.py <path_to_server_script>")
262 sys.exit(1)
263
264 client = MCPClient()
202 "content": [
203 {
204 "type": "tool_result",
205 "tool_use_id": content.id,
206 "content": result.content
207 }
208 ]
209 })
210
211 # Get next response from Claude
212 response = self.anthropic.messages.create(
213 model="claude-sonnet-4-20250514",
214 max_tokens=1000,
215 messages=messages,
216 tools=available_tools
217 )
218
219 final_text.append(response.content[0].text)
220
221 return "\n".join(final_text)
222```
223
224### Interactive Chat Interface
225
226Now we'll add the chat loop and cleanup functionality:
227
228```python theme={null}
229async def chat_loop(self):
230 """Run an interactive chat loop"""
231 print("\nMCP Client Started!")
232 print("Type your queries or 'quit' to exit.")
233
234 while True:
265235 try:
266 await client.connect_to_server(sys.argv[1])
267 await client.chat_loop()
268 finally:
269 await client.cleanup()
270
271 if __name__ == "__main__":
272 import sys
273 asyncio.run(main())
274 ```
275
276 You can find the complete `client.py` file [here](https://github.com/modelcontextprotocol/quickstart-resources/blob/main/mcp-client-python/client.py).
277
278 ## Key Components Explained
279
280 ### 1. Client Initialization
281
282 * The `MCPClient` class initializes with session management and API clients
283 * Uses `AsyncExitStack` for proper resource management
284 * Configures the Anthropic client for Claude interactions
285
286 ### 2. Server Connection
287
288 * Supports both Python and Node.js servers
289 * Validates server script type
290 * Sets up proper communication channels
291 * Initializes the session and lists available tools
292
293 ### 3. Query Processing
294
295 * Maintains conversation context
296 * Handles Claude's responses and tool calls
297 * Manages the message flow between Claude and tools
298 * Combines results into a coherent response
299
300 ### 4. Interactive Interface
301
302 * Provides a simple command-line interface
303 * Handles user input and displays responses
304 * Includes basic error handling
305 * Allows graceful exit
306
307 ### 5. Resource Management
308
309 * Proper cleanup of resources
310 * Error handling for connection issues
311 * Graceful shutdown procedures
312
313 ## Common Customization Points
314
315 1. **Tool Handling**
316 * Modify `process_query()` to handle specific tool types
317 * Add custom error handling for tool calls
318 * Implement tool-specific response formatting
319
320 2. **Response Processing**
321 * Customize how tool results are formatted
322 * Add response filtering or transformation
323 * Implement custom logging
324
325 3. **User Interface**
326 * Add a GUI or web interface
327 * Implement rich console output
328 * Add command history or auto-completion
329
330 ## Running the Client
331
332 To run your client with any MCP server:
333
334 ```bash theme={null}
335 uv run client.py path/to/server.py # python server
336 uv run client.py path/to/build/index.js # node server
337 ```
338
339 <Note>
340 If you're continuing [the weather tutorial from the server quickstart](https://github.com/modelcontextprotocol/quickstart-resources/tree/main/weather-server-python), your command might look something like this: `python client.py .../quickstart-resources/weather-server-python/weather.py`
341 </Note>
342
343 The client will:
344
345 1. Connect to the specified server
346 2. List available tools
347 3. Start an interactive chat session where you can:
348 * Enter queries
349 * See tool executions
350 * Get responses from Claude
351
352 Here's an example of what it should look like if connected to the weather server from the server quickstart:
353
354 <Frame>
355 <img src="https://mintcdn.com/mcp/4ZXF1PrDkEaJvXpn/images/client-claude-cli-python.png?fit=max&auto=format&n=4ZXF1PrDkEaJvXpn&q=85&s=686d6e0ae7c54f807827db111eaed7d4" width="1932" height="1739" data-path="images/client-claude-cli-python.png" />
356 </Frame>
357
358 ## How It Works
359
360 When you submit a query:
361
362 1. The client gets the list of available tools from the server
363 2. Your query is sent to Claude along with tool descriptions
364 3. Claude decides which tools (if any) to use
365 4. The client executes any requested tool calls through the server
366 5. Results are sent back to Claude
367 6. Claude provides a natural language response
368 7. The response is displayed to you
369
370 ## Best practices
371
372 1. **Error Handling**
373 * Always wrap tool calls in try-catch blocks
374 * Provide meaningful error messages
375 * Gracefully handle connection issues
376
377 2. **Resource Management**
378 * Use `AsyncExitStack` for proper cleanup
379 * Close connections when done
380 * Handle server disconnections
381
382 3. **Security**
383 * Store API keys securely in `.env`
384 * Validate server responses
385 * Be cautious with tool permissions
386
387 ## Troubleshooting
388
389 ### Server Path Issues
390
391 * Double-check the path to your server script is correct
392 * Use the absolute path if the relative path isn't working
393 * For Windows users, make sure to use forward slashes (/) or escaped backslashes (\\) in the path
394 * Verify the server file has the correct extension (.py for Python or .js for Node.js)
395
396 Example of correct path usage:
397
398 ```bash theme={null}
399 # Relative path
400 uv run client.py ./server/weather.py
401
402 # Absolute path
403 uv run client.py /Users/username/projects/mcp-server/weather.py
404
405 # Windows path (either format works)
406 uv run client.py C:/projects/mcp-server/weather.py
407 uv run client.py C:\\projects\\mcp-server\\weather.py
408 ```
409
410 ### Response Timing
411
412 * The first response might take up to 30 seconds to return
413 * This is normal and happens while:
414 * The server initializes
415 * Claude processes the query
416 * Tools are being executed
417 * Subsequent responses are typically faster
418 * Don't interrupt the process during this initial waiting period
419
420 ### Common Error Messages
421
422 If you see:
423
424 * `FileNotFoundError`: Check your server path
425 * `Connection refused`: Ensure the server is running and the path is correct
426 * `Tool execution failed`: Verify the tool's required environment variables are set
427 * `Timeout error`: Consider increasing the timeout in your client configuration
428 </Tab>
429
430 <Tab title="TypeScript">
431 [You can find the complete code for this tutorial here.](https://github.com/modelcontextprotocol/quickstart-resources/tree/main/mcp-client-typescript)
432
433 ## System Requirements
434
435 Before starting, ensure your system meets these requirements:
436
437 * Mac or Windows computer
438 * Node.js 17 or higher installed
439 * Latest version of `npm` installed
440 * Anthropic API key (Claude)
441
442 ## Setting Up Your Environment
443
444 First, let's create and set up our project:
445
446 <CodeGroup>
447 ```bash macOS/Linux theme={null}
448 # Create project directory
449 mkdir mcp-client-typescript
450 cd mcp-client-typescript
451
452 # Initialize npm project
453 npm init -y
454
455 # Install dependencies
456 npm install @anthropic-ai/sdk @modelcontextprotocol/sdk dotenv
457
458 # Install dev dependencies
459 npm install -D @types/node typescript
460
461 # Create source file
462 touch index.ts
463 ```
464
465 ```powershell Windows theme={null}
466 # Create project directory
467 md mcp-client-typescript
468 cd mcp-client-typescript
469
470 # Initialize npm project
471 npm init -y
472
473 # Install dependencies
474 npm install @anthropic-ai/sdk @modelcontextprotocol/sdk dotenv
475
476 # Install dev dependencies
477 npm install -D @types/node typescript
478
479 # Create source file
480 new-item index.ts
481 ```
482 </CodeGroup>
483
484 Update your `package.json` to set `type: "module"` and a build script:
485
486 ```json package.json theme={null}
236 query = input("\nQuery: ").strip()
237
238 if query.lower() == 'quit':
239 break
240
241 response = await self.process_query(query)
242 print("\n" + response)
243
244 except Exception as e:
245 print(f"\nError: {str(e)}")
246
247async def cleanup(self):
248 """Clean up resources"""
249 await self.exit_stack.aclose()
250```
251
252### Main Entry Point
253
254Finally, we'll add the main execution logic:
255
256```python theme={null}
257async def main():
258 if len(sys.argv) < 2:
259 print("Usage: python client.py <path_to_server_script>")
260 sys.exit(1)
261
262 client = MCPClient()
263 try:
264 await client.connect_to_server(sys.argv[1])
265 await client.chat_loop()
266 finally:
267 await client.cleanup()
268
269if __name__ == "__main__":
270 import sys
271 asyncio.run(main())
272```
273
274You can find the complete `client.py` file [here](https://github.com/modelcontextprotocol/quickstart-resources/blob/main/mcp-client-python/client.py).
275
276## Key Components Explained
277
278### 1. Client Initialization
279
280* The `MCPClient` class initializes with session management and API clients
281* Uses `AsyncExitStack` for proper resource management
282* Configures the Anthropic client for Claude interactions
283
284### 2. Server Connection
285
286* Supports both Python and Node.js servers
287* Validates server script type
288* Sets up proper communication channels
289* Initializes the session and lists available tools
290
291### 3. Query Processing
292
293* Maintains conversation context
294* Handles Claude's responses and tool calls
295* Manages the message flow between Claude and tools
296* Combines results into a coherent response
297
298### 4. Interactive Interface
299
300* Provides a simple command-line interface
301* Handles user input and displays responses
302* Includes basic error handling
303* Allows graceful exit
304
305### 5. Resource Management
306
307* Proper cleanup of resources
308* Error handling for connection issues
309* Graceful shutdown procedures
310
311## Common Customization Points
312
3131. **Tool Handling**
314 * Modify `process_query()` to handle specific tool types
315 * Add custom error handling for tool calls
316 * Implement tool-specific response formatting
317
3182. **Response Processing**
319 * Customize how tool results are formatted
320 * Add response filtering or transformation
321 * Implement custom logging
322
3233. **User Interface**
324 * Add a GUI or web interface
325 * Implement rich console output
326 * Add command history or auto-completion
327
328## Running the Client
329
330To run your client with any MCP server:
331
332```bash theme={null}
333uv run client.py path/to/server.py # python server
334uv run client.py path/to/build/index.js # node server
335```
336
337> [!NOTE]
338>
339> If you're continuing [the weather tutorial from the server quickstart](https://github.com/modelcontextprotocol/quickstart-resources/tree/main/weather-server-python), your command might look something like this: `python client.py .../quickstart-resources/weather-server-python/weather.py`
340
341The client will:
342
3431. Connect to the specified server
3442. List available tools
3453. Start an interactive chat session where you can:
346 * Enter queries
347 * See tool executions
348 * Get responses from Claude
349
350Here's an example of what it should look like if connected to the weather server from the server quickstart:
351
352
353
354## How It Works
355
356When you submit a query:
357
3581. The client gets the list of available tools from the server
3592. Your query is sent to Claude along with tool descriptions
3603. Claude decides which tools (if any) to use
3614. The client executes any requested tool calls through the server
3625. Results are sent back to Claude
3636. Claude provides a natural language response
3647. The response is displayed to you
365
366## Best practices
367
3681. **Error Handling**
369 * Always wrap tool calls in try-catch blocks
370 * Provide meaningful error messages
371 * Gracefully handle connection issues
372
3732. **Resource Management**
374 * Use `AsyncExitStack` for proper cleanup
375 * Close connections when done
376 * Handle server disconnections
377
3783. **Security**
379 * Store API keys securely in `.env`
380 * Validate server responses
381 * Be cautious with tool permissions
382
383## Troubleshooting
384
385### Server Path Issues
386
387* Double-check the path to your server script is correct
388* Use the absolute path if the relative path isn't working
389* For Windows users, make sure to use forward slashes (/) or escaped backslashes (\\) in the path
390* Verify the server file has the correct extension (.py for Python or .js for Node.js)
391
392Example of correct path usage:
393
394```bash theme={null}
395# Relative path
396uv run client.py ./server/weather.py
397
398# Absolute path
399uv run client.py /Users/username/projects/mcp-server/weather.py
400
401# Windows path (either format works)
402uv run client.py C:/projects/mcp-server/weather.py
403uv run client.py C:\\projects\\mcp-server\\weather.py
404```
405
406### Response Timing
407
408* The first response might take up to 30 seconds to return
409* This is normal and happens while:
410 * The server initializes
411 * Claude processes the query
412 * Tools are being executed
413* Subsequent responses are typically faster
414* Don't interrupt the process during this initial waiting period
415
416### Common Error Messages
417
418If you see:
419
420* `FileNotFoundError`: Check your server path
421* `Connection refused`: Ensure the server is running and the path is correct
422* `Tool execution failed`: Verify the tool's required environment variables are set
423* `Timeout error`: Consider increasing the timeout in your client configuration
424
425**TypeScript**
426
427[You can find the complete code for this tutorial here.](https://github.com/modelcontextprotocol/quickstart-resources/tree/main/mcp-client-typescript)
428
429## System Requirements
430
431Before starting, ensure your system meets these requirements:
432
433* Mac or Windows computer
434* Node.js 17 or higher installed
435* Latest version of `npm` installed
436* Anthropic API key (Claude)
437
438## Setting Up Your Environment
439
440First, let's create and set up our project:
441
442```bash macOS/Linux theme={null}
443# Create project directory
444mkdir mcp-client-typescript
445cd mcp-client-typescript
446
447# Initialize npm project
448npm init -y
449
450# Install dependencies
451npm install @anthropic-ai/sdk @modelcontextprotocol/sdk dotenv
452
453# Install dev dependencies
454npm install -D @types/node typescript
455
456# Create source file
457touch index.ts
458```
459
460```powershell Windows theme={null}
461# Create project directory
462md mcp-client-typescript
463cd mcp-client-typescript
464
465# Initialize npm project
466npm init -y
467
468# Install dependencies
469npm install @anthropic-ai/sdk @modelcontextprotocol/sdk dotenv
470
471# Install dev dependencies
472npm install -D @types/node typescript
473
474# Create source file
475new-item index.ts
476```
477
478Update your `package.json` to set `type: "module"` and a build script:
479
480```json package.json theme={null}
481{
482 "type": "module",
483 "scripts": {
484 "build": "tsc && chmod 755 build/index.js"
485 }
486}
487```
488
489Create a `tsconfig.json` in the root of your project:
490
491```json tsconfig.json theme={null}
492{
493 "compilerOptions": {
494 "target": "ES2022",
495 "module": "Node16",
496 "moduleResolution": "Node16",
497 "outDir": "./build",
498 "rootDir": "./",
499 "strict": true,
500 "esModuleInterop": true,
501 "skipLibCheck": true,
502 "forceConsistentCasingInFileNames": true
503 },
504 "include": ["index.ts"],
505 "exclude": ["node_modules"]
506}
507```
508
509## Setting Up Your API Key
510
511You'll need an Anthropic API key from the [Anthropic Console](https://console.anthropic.com/settings/keys).
512
513Create a `.env` file to store it:
514
515```bash theme={null}
516echo "ANTHROPIC_API_KEY=<your key here>" > .env
517```
518
519Add `.env` to your `.gitignore`:
520
521```bash theme={null}
522echo ".env" >> .gitignore
523```
524
525> [!WARNING]
526>
527> Make sure you keep your `ANTHROPIC_API_KEY` secure!
528
529## Creating the Client
530
531### Basic Client Structure
532
533First, let's set up our imports and create the basic client class in `index.ts`:
534
535```typescript theme={null}
536import { Anthropic } from "@anthropic-ai/sdk";
537import {
538 MessageParam,
539 Tool,
540} from "@anthropic-ai/sdk/resources/messages/messages.mjs";
541import { Client } from "@modelcontextprotocol/sdk/client/index.js";
542import { StdioClientTransport } from "@modelcontextprotocol/sdk/client/stdio.js";
543import readline from "readline/promises";
544import dotenv from "dotenv";
545
546dotenv.config();
547
548const ANTHROPIC_API_KEY = process.env.ANTHROPIC_API_KEY;
549if (!ANTHROPIC_API_KEY) {
550 throw new Error("ANTHROPIC_API_KEY is not set");
551}
552
553class MCPClient {
554 private mcp: Client;
555 private anthropic: Anthropic;
556 private transport: StdioClientTransport | null = null;
557 private tools: Tool[] = [];
558
559 constructor() {
560 this.anthropic = new Anthropic({
561 apiKey: ANTHROPIC_API_KEY,
562 });
563 this.mcp = new Client({ name: "mcp-client-cli", version: "1.0.0" });
564 }
565 // methods will go here
566}
567```
568
569### Server Connection Management
570
571Next, we'll implement the method to connect to an MCP server:
572
573```typescript theme={null}
574async connectToServer(serverScriptPath: string) {
575 try {
576 const isJs = serverScriptPath.endsWith(".js");
577 const isPy = serverScriptPath.endsWith(".py");
578 if (!isJs && !isPy) {
579 throw new Error("Server script must be a .js or .py file");
580 }
581 const command = isPy
582 ? process.platform === "win32"
583 ? "python"
584 : "python3"
585 : process.execPath;
586
587 this.transport = new StdioClientTransport({
588 command,
589 args: [serverScriptPath],
590 });
591 await this.mcp.connect(this.transport);
592
593 const toolsResult = await this.mcp.listTools();
594 this.tools = toolsResult.tools.map((tool) => {
595 return {
596 name: tool.name,
597 description: tool.description,
598 input_schema: tool.inputSchema,
599 };
600 });
601 console.log(
602 "Connected to server with tools:",
603 this.tools.map(({ name }) => name)
604 );
605 } catch (e) {
606 console.log("Failed to connect to MCP server: ", e);
607 throw e;
608 }
609}
610```
611
612### Query Processing Logic
613
614Now let's add the core functionality for processing queries and handling tool calls:
615
616```typescript theme={null}
617async processQuery(query: string) {
618 const messages: MessageParam[] = [
487619 {
488 "type": "module",
489 "scripts": {
490 "build": "tsc && chmod 755 build/index.js"
491 }
492 }
493 ```
494
495 Create a `tsconfig.json` in the root of your project:
496
497 ```json tsconfig.json theme={null}
498 {
499 "compilerOptions": {
500 "target": "ES2022",
501 "module": "Node16",
502 "moduleResolution": "Node16",
503 "outDir": "./build",
504 "rootDir": "./",
505 "strict": true,
506 "esModuleInterop": true,
507 "skipLibCheck": true,
508 "forceConsistentCasingInFileNames": true
509 },
510 "include": ["index.ts"],
511 "exclude": ["node_modules"]
512 }
513 ```
514
515 ## Setting Up Your API Key
516
517 You'll need an Anthropic API key from the [Anthropic Console](https://console.anthropic.com/settings/keys).
518
519 Create a `.env` file to store it:
520
521 ```bash theme={null}
522 echo "ANTHROPIC_API_KEY=<your key here>" > .env
523 ```
524
525 Add `.env` to your `.gitignore`:
526
527 ```bash theme={null}
528 echo ".env" >> .gitignore
529 ```
530
531 <Warning>
532 Make sure you keep your `ANTHROPIC_API_KEY` secure!
533 </Warning>
534
535 ## Creating the Client
536
537 ### Basic Client Structure
538
539 First, let's set up our imports and create the basic client class in `index.ts`:
540
541 ```typescript theme={null}
542 import { Anthropic } from "@anthropic-ai/sdk";
543 import {
544 MessageParam,
545 Tool,
546 } from "@anthropic-ai/sdk/resources/messages/messages.mjs";
547 import { Client } from "@modelcontextprotocol/sdk/client/index.js";
548 import { StdioClientTransport } from "@modelcontextprotocol/sdk/client/stdio.js";
549 import readline from "readline/promises";
550 import dotenv from "dotenv";
551
552 dotenv.config();
553
554 const ANTHROPIC_API_KEY = process.env.ANTHROPIC_API_KEY;
555 if (!ANTHROPIC_API_KEY) {
556 throw new Error("ANTHROPIC_API_KEY is not set");
557 }
558
559 class MCPClient {
560 private mcp: Client;
561 private anthropic: Anthropic;
562 private transport: StdioClientTransport | null = null;
563 private tools: Tool[] = [];
564
565 constructor() {
566 this.anthropic = new Anthropic({
567 apiKey: ANTHROPIC_API_KEY,
568 });
569 this.mcp = new Client({ name: "mcp-client-cli", version: "1.0.0" });
570 }
571 // methods will go here
572 }
573 ```
574
575 ### Server Connection Management
576
577 Next, we'll implement the method to connect to an MCP server:
578
579 ```typescript theme={null}
580 async connectToServer(serverScriptPath: string) {
581 try {
582 const isJs = serverScriptPath.endsWith(".js");
583 const isPy = serverScriptPath.endsWith(".py");
584 if (!isJs && !isPy) {
585 throw new Error("Server script must be a .js or .py file");
586 }
587 const command = isPy
588 ? process.platform === "win32"
589 ? "python"
590 : "python3"
591 : process.execPath;
592
593 this.transport = new StdioClientTransport({
594 command,
595 args: [serverScriptPath],
596 });
597 await this.mcp.connect(this.transport);
598
599 const toolsResult = await this.mcp.listTools();
600 this.tools = toolsResult.tools.map((tool) => {
601 return {
602 name: tool.name,
603 description: tool.description,
604 input_schema: tool.inputSchema,
605 };
606 });
607 console.log(
608 "Connected to server with tools:",
609 this.tools.map(({ name }) => name)
610 );
611 } catch (e) {
612 console.log("Failed to connect to MCP server: ", e);
613 throw e;
614 }
615 }
616 ```
617
618 ### Query Processing Logic
619
620 Now let's add the core functionality for processing queries and handling tool calls:
621
622 ```typescript theme={null}
623 async processQuery(query: string) {
624 const messages: MessageParam[] = [
625 {
626 role: "user",
627 content: query,
628 },
629 ];
620 role: "user",
621 content: query,
622 },
623 ];
624
625 const response = await this.anthropic.messages.create({
626 model: "claude-sonnet-4-20250514",
627 max_tokens: 1000,
628 messages,
629 tools: this.tools,
630 });
631
632 const finalText = [];
633
634 for (const content of response.content) {
635 if (content.type === "text") {
636 finalText.push(content.text);
637 } else if (content.type === "tool_use") {
638 const toolName = content.name;
639 const toolArgs = content.input as { [x: string]: unknown } | undefined;
640
641 const result = await this.mcp.callTool({
642 name: toolName,
643 arguments: toolArgs,
644 });
645 finalText.push(
646 `[Calling tool ${toolName} with args ${JSON.stringify(toolArgs)}]`
647 );
648
649 messages.push({
650 role: "user",
651 content: result.content as string,
652 });
630653
631654 const response = await this.anthropic.messages.create({
632655 model: "claude-sonnet-4-20250514",
633656 max_tokens: 1000,
634657 messages,
635 tools: this.tools,
636658 });
637659
638 const finalText = [];
639
640 for (const content of response.content) {
641 if (content.type === "text") {
642 finalText.push(content.text);
643 } else if (content.type === "tool_use") {
644 const toolName = content.name;
645 const toolArgs = content.input as { [x: string]: unknown } | undefined;
646
647 const result = await this.mcp.callTool({
648 name: toolName,
649 arguments: toolArgs,
650 });
651 finalText.push(
652 `[Calling tool ${toolName} with args ${JSON.stringify(toolArgs)}]`
653 );
654
655 messages.push({
656 role: "user",
657 content: result.content as string,
658 });
659
660 const response = await this.anthropic.messages.create({
661 model: "claude-sonnet-4-20250514",
662 max_tokens: 1000,
663 messages,
664 });
665
666 finalText.push(
667 response.content[0].type === "text" ? response.content[0].text : ""
668 );
669 }
660 finalText.push(
661 response.content[0].type === "text" ? response.content[0].text : ""
662 );
663 }
664 }
665
666 return finalText.join("\n");
667}
668```
669
670### Interactive Chat Interface
671
672Now we'll add the chat loop and cleanup functionality:
673
674```typescript theme={null}
675async chatLoop() {
676 const rl = readline.createInterface({
677 input: process.stdin,
678 output: process.stdout,
679 });
680
681 try {
682 console.log("\nMCP Client Started!");
683 console.log("Type your queries or 'quit' to exit.");
684
685 while (true) {
686 const message = await rl.question("\nQuery: ");
687 if (message.toLowerCase() === "quit") {
688 break;
670689 }
671
672 return finalText.join("\n");
690 const response = await this.processQuery(message);
691 console.log("\n" + response);
673692 }
674 ```
675
676 ### Interactive Chat Interface
677
678 Now we'll add the chat loop and cleanup functionality:
679
680 ```typescript theme={null}
681 async chatLoop() {
682 const rl = readline.createInterface({
683 input: process.stdin,
684 output: process.stdout,
685 });
686
687 try {
688 console.log("\nMCP Client Started!");
689 console.log("Type your queries or 'quit' to exit.");
690
691 while (true) {
692 const message = await rl.question("\nQuery: ");
693 if (message.toLowerCase() === "quit") {
694 break;
695 }
696 const response = await this.processQuery(message);
697 console.log("\n" + response);
698 }
699 } finally {
700 rl.close();
693 } finally {
694 rl.close();
695 }
696}
697
698async cleanup() {
699 await this.mcp.close();
700}
701```
702
703### Main Entry Point
704
705Finally, we'll add the main execution logic:
706
707```typescript theme={null}
708async function main() {
709 if (process.argv.length < 3) {
710 console.log("Usage: node index.ts <path_to_server_script>");
711 return;
712 }
713 const mcpClient = new MCPClient();
714 try {
715 await mcpClient.connectToServer(process.argv[2]);
716 await mcpClient.chatLoop();
717 } catch (e) {
718 console.error("Error:", e);
719 await mcpClient.cleanup();
720 process.exit(1);
721 } finally {
722 await mcpClient.cleanup();
723 process.exit(0);
724 }
725}
726
727main();
728```
729
730## Running the Client
731
732To run your client with any MCP server:
733
734```bash theme={null}
735# Build TypeScript
736npm run build
737
738# Run the client
739node build/index.js path/to/server.py # python server
740node build/index.js path/to/build/index.js # node server
741```
742
743> [!NOTE]
744>
745> If you're continuing [the weather tutorial from the server quickstart](https://github.com/modelcontextprotocol/quickstart-resources/tree/main/weather-server-typescript), your command might look something like this: `node build/index.js .../quickstart-resources/weather-server-typescript/build/index.js`
746
747**The client will:**
748
7491. Connect to the specified server
7502. List available tools
7513. Start an interactive chat session where you can:
752 * Enter queries
753 * See tool executions
754 * Get responses from Claude
755
756## How It Works
757
758When you submit a query:
759
7601. The client gets the list of available tools from the server
7612. Your query is sent to Claude along with tool descriptions
7623. Claude decides which tools (if any) to use
7634. The client executes any requested tool calls through the server
7645. Results are sent back to Claude
7656. Claude provides a natural language response
7667. The response is displayed to you
767
768## Best practices
769
7701. **Error Handling**
771 * Use TypeScript's type system for better error detection
772 * Wrap tool calls in try-catch blocks
773 * Provide meaningful error messages
774 * Gracefully handle connection issues
775
7762. **Security**
777 * Store API keys securely in `.env`
778 * Validate server responses
779 * Be cautious with tool permissions
780
781## Troubleshooting
782
783### Server Path Issues
784
785* Double-check the path to your server script is correct
786* Use the absolute path if the relative path isn't working
787* For Windows users, make sure to use forward slashes (/) or escaped backslashes (\\) in the path
788* Verify the server file has the correct extension (.js for Node.js or .py for Python)
789
790Example of correct path usage:
791
792```bash theme={null}
793# Relative path
794node build/index.js ./server/build/index.js
795
796# Absolute path
797node build/index.js /Users/username/projects/mcp-server/build/index.js
798
799# Windows path (either format works)
800node build/index.js C:/projects/mcp-server/build/index.js
801node build/index.js C:\\projects\\mcp-server\\build\\index.js
802```
803
804### Response Timing
805
806* The first response might take up to 30 seconds to return
807* This is normal and happens while:
808 * The server initializes
809 * Claude processes the query
810 * Tools are being executed
811* Subsequent responses are typically faster
812* Don't interrupt the process during this initial waiting period
813
814### Common Error Messages
815
816If you see:
817
818* `Error: Cannot find module`: Check your build folder and ensure TypeScript compilation succeeded
819* `Connection refused`: Ensure the server is running and the path is correct
820* `Tool execution failed`: Verify the tool's required environment variables are set
821* `ANTHROPIC_API_KEY is not set`: Check your .env file and environment variables
822* `TypeError`: Ensure you're using the correct types for tool arguments
823* `BadRequestError`: Ensure you have enough credits to access the Anthropic API
824
825**Java**
826
827> [!NOTE]
828>
829> This is a quickstart demo based on Spring AI MCP auto-configuration and boot starters.
830> To learn how to create sync and async MCP Clients manually, consult the [Java SDK Client](https://java.sdk.modelcontextprotocol.io/) documentation
831
832This example demonstrates how to build an interactive chatbot that combines Spring AI's Model Context Protocol (MCP) with the [Brave Search MCP Server](https://github.com/modelcontextprotocol/servers-archived/tree/main/src/brave-search). The application creates a conversational interface powered by Anthropic's Claude AI model that can perform internet searches through Brave Search, enabling natural language interactions with real-time web data.
833[You can find the complete code for this tutorial here.](https://github.com/spring-projects/spring-ai-examples/tree/main/model-context-protocol/web-search/brave-chatbot)
834
835## System Requirements
836
837Before starting, ensure your system meets these requirements:
838
839* Java 17 or higher
840* Maven 3.6+
841* npx package manager
842* Anthropic API key (Claude)
843* Brave Search API key
844
845## Setting Up Your Environment
846
8471. Install npx (Node Package eXecute):
848 First, make sure to install [npm](https://docs.npmjs.com/downloading-and-installing-node-js-and-npm)
849 and then run:
850
851 ```bash theme={null}
852 npm install -g npx
853 ```
854
8552. Clone the repository:
856
857 ```bash theme={null}
858 git clone https://github.com/spring-projects/spring-ai-examples.git
859 cd model-context-protocol/web-search/brave-chatbot
860 ```
861
8623. Set up your API keys:
863
864 ```bash theme={null}
865 export ANTHROPIC_API_KEY='your-anthropic-api-key-here'
866 export BRAVE_API_KEY='your-brave-api-key-here'
867 ```
868
8694. Build the application:
870
871 ```bash theme={null}
872 ./mvnw clean install
873 ```
874
8755. Run the application using Maven:
876 ```bash theme={null}
877 ./mvnw spring-boot:run
878 ```
879
880> [!WARNING]
881>
882> Make sure you keep your `ANTHROPIC_API_KEY` and `BRAVE_API_KEY` keys secure!
883
884## How it Works
885
886The application integrates Spring AI with the Brave Search MCP server through several components:
887
888### MCP Client Configuration
889
8901. Required dependencies in pom.xml:
891
892```xml theme={null}
893<dependency>
894 <groupId>org.springframework.ai</groupId>
895 <artifactId>spring-ai-starter-mcp-client</artifactId>
896</dependency>
897<dependency>
898 <groupId>org.springframework.ai</groupId>
899 <artifactId>spring-ai-starter-model-anthropic</artifactId>
900</dependency>
901```
902
9032. Application properties (application.yml):
904
905```yml theme={null}
906spring:
907 ai:
908 mcp:
909 client:
910 enabled: true
911 name: brave-search-client
912 version: 1.0.0
913 type: SYNC
914 request-timeout: 20s
915 stdio:
916 root-change-notification: true
917 servers-configuration: classpath:/mcp-servers-config.json
918 toolcallback:
919 enabled: true
920 anthropic:
921 api-key: ${ANTHROPIC_API_KEY}
922```
923
924This activates the `spring-ai-starter-mcp-client` to create one or more `McpClient`s based on the provided server configuration.
925The `spring.ai.mcp.client.toolcallback.enabled=true` property enables the tool callback mechanism, that automatically registers all MCP tool as spring ai tools.
926It is disabled by default.
927
9283. MCP Server Configuration (`mcp-servers-config.json`):
929
930```json theme={null}
931{
932 "mcpServers": {
933 "brave-search": {
934 "command": "npx",
935 "args": ["-y", "@modelcontextprotocol/server-brave-search"],
936 "env": {
937 "BRAVE_API_KEY": "<PUT YOUR BRAVE API KEY>"
701938 }
702939 }
703
704 async cleanup() {
705 await this.mcp.close();
940 }
941}
942```
943
944### Chat Implementation
945
946The chatbot is implemented using Spring AI's ChatClient with MCP tool integration:
947
948```java theme={null}
949var chatClient = chatClientBuilder
950 .defaultSystem("You are useful assistant, expert in AI and Java.")
951 .defaultToolCallbacks((Object[]) mcpToolAdapter.toolCallbacks())
952 .defaultAdvisors(new MessageChatMemoryAdvisor(new InMemoryChatMemory()))
953 .build();
954```
955
956Key features:
957
958* Uses Claude AI model for natural language understanding
959* Integrates Brave Search through MCP for real-time web search capabilities
960* Maintains conversation memory using InMemoryChatMemory
961* Runs as an interactive command-line application
962
963### Build and run
964
965```bash theme={null}
966./mvnw clean install
967java -jar ./target/ai-mcp-brave-chatbot-0.0.1-SNAPSHOT.jar
968```
969
970or
971
972```bash theme={null}
973./mvnw spring-boot:run
974```
975
976The application will start an interactive chat session where you can ask questions. The chatbot will use Brave Search when it needs to find information from the internet to answer your queries.
977
978The chatbot can:
979
980* Answer questions using its built-in knowledge
981* Perform web searches when needed using Brave Search
982* Remember context from previous messages in the conversation
983* Combine information from multiple sources to provide comprehensive answers
984
985### Advanced Configuration
986
987The MCP client supports additional configuration options:
988
989* Client customization through `McpSyncClientCustomizer` or `McpAsyncClientCustomizer`
990* Multiple clients with multiple transport types: `STDIO` and `SSE` (Server-Sent Events)
991* Integration with Spring AI's tool execution framework
992* Automatic client initialization and lifecycle management
993
994For WebFlux-based applications, you can use the WebFlux starter instead:
995
996```xml theme={null}
997<dependency>
998 <groupId>org.springframework.ai</groupId>
999 <artifactId>spring-ai-mcp-client-webflux-spring-boot-starter</artifactId>
1000</dependency>
1001```
1002
1003This provides similar functionality but uses a WebFlux-based SSE transport implementation, recommended for production deployments.
1004
1005**Kotlin**
1006
1007[You can find the complete code for this tutorial here.](https://github.com/modelcontextprotocol/kotlin-sdk/tree/main/samples/kotlin-mcp-client)
1008
1009## System Requirements
1010
1011Before starting, ensure your system meets these requirements:
1012
1013* JDK 11 or higher
1014* Anthropic API key (Claude)
1015
1016## Setting up your environment
1017
1018First, let's install `java` and `gradle` if you haven't already.
1019You can download `java` from [official Oracle JDK website](https://www.oracle.com/java/technologies/downloads/).
1020Verify your `java` installation:
1021
1022```bash theme={null}
1023java --version
1024```
1025
1026Now, let's create and set up your project:
1027
1028```bash macOS/Linux theme={null}
1029# Create a new directory for our project
1030mkdir kotlin-mcp-client
1031cd kotlin-mcp-client
1032
1033# Initialize a new kotlin project
1034gradle init
1035```
1036
1037```powershell Windows theme={null}
1038# Create a new directory for our project
1039md kotlin-mcp-client
1040cd kotlin-mcp-client
1041# Initialize a new kotlin project
1042gradle init
1043```
1044
1045After running `gradle init`, select **Application** as the project type, **Kotlin** as the programming language.
1046
1047Alternatively, you can create a Kotlin application using the [IntelliJ IDEA project wizard](https://kotlinlang.org/docs/jvm-get-started.html).
1048
1049After creating the project, replace the contents of your `build.gradle.kts` with:
1050
1051```kotlin build.gradle.kts theme={null}
1052// Check latest versions at https://github.com/modelcontextprotocol/kotlin-sdk/releases
1053val mcpVersion = "0.9.0"
1054val ktorVersion = "3.2.3"
1055val anthropicVersion = "2.15.0"
1056val slf4jVersion = "2.0.17"
1057
1058plugins {
1059 kotlin("jvm") version "2.3.20"
1060 id("com.gradleup.shadow") version "8.3.9"
1061 application
1062}
1063
1064application {
1065 mainClass.set("MainKt")
1066}
1067
1068dependencies {
1069 implementation("io.modelcontextprotocol:kotlin-sdk:$mcpVersion")
1070 implementation("io.ktor:ktor-client-cio:$ktorVersion")
1071 implementation("com.anthropic:anthropic-java:$anthropicVersion")
1072 implementation("org.slf4j:slf4j-simple:$slf4jVersion")
1073}
1074```
1075
1076Verify that everything is set up correctly:
1077
1078```bash theme={null}
1079./gradlew build
1080```
1081
1082## Setting up your API key
1083
1084You'll need an Anthropic API key from the [Anthropic Console](https://console.anthropic.com/settings/keys).
1085
1086Set up your API key:
1087
1088```bash theme={null}
1089export ANTHROPIC_API_KEY='your-anthropic-api-key-here'
1090```
1091
1092> [!WARNING]
1093>
1094> Make sure you keep your `ANTHROPIC_API_KEY` secure!
1095
1096## Creating the Client
1097
1098### Basic Client Structure
1099
1100First, let's create the basic client class:
1101
1102```kotlin theme={null}
1103class MCPClient(apiKey: String) : AutoCloseable {
1104 private val anthropic = AnthropicOkHttpClient.builder()
1105 .apiKey(apiKey)
1106 .build()
1107
1108 private val mcp: Client = Client(
1109 clientInfo = Implementation(name = "mcp-client-cli", version = "1.0.0")
1110 )
1111 private var serverProcess: Process? = null
1112 private lateinit var tools: List<ToolUnion>
1113
1114 // methods will go here
1115
1116 override fun close() {
1117 runBlocking {
1118 mcp.close()
1119 }
1120 serverProcess?.destroy()
1121 anthropic.close()
7061122 }
707 ```
708
709 ### Main Entry Point
710
711 Finally, we'll add the main execution logic:
712
713 ```typescript theme={null}
714 async function main() {
715 if (process.argv.length < 3) {
716 console.log("Usage: node index.ts <path_to_server_script>");
717 return;
718 }
719 const mcpClient = new MCPClient();
720 try {
721 await mcpClient.connectToServer(process.argv[2]);
722 await mcpClient.chatLoop();
723 } catch (e) {
724 console.error("Error:", e);
725 await mcpClient.cleanup();
726 process.exit(1);
727 } finally {
728 await mcpClient.cleanup();
729 process.exit(0);
730 }
1123}
1124```
1125
1126### Server connection management
1127
1128Next, we'll implement the method to connect to an MCP server:
1129
1130```kotlin theme={null}
1131suspend fun connectToServer(serverScriptPath: String) {
1132 val command = buildList {
1133 when (serverScriptPath.substringAfterLast(".")) {
1134 "js" -> add("node")
1135 "py" -> add(if (System.getProperty("os.name").lowercase().contains("win")) "python" else "python3")
1136 "jar" -> addAll(listOf("java", "-jar"))
1137 else -> throw IllegalArgumentException("Server script must be a .js, .py or .jar file")
1138 }
1139 add(serverScriptPath)
7311140 }
7321141
733 main();
734 ```
735
736 ## Running the Client
737
738 To run your client with any MCP server:
739
740 ```bash theme={null}
741 # Build TypeScript
742 npm run build
743
744 # Run the client
745 node build/index.js path/to/server.py # python server
746 node build/index.js path/to/build/index.js # node server
747 ```
748
749 <Note>
750 If you're continuing [the weather tutorial from the server quickstart](https://github.com/modelcontextprotocol/quickstart-resources/tree/main/weather-server-typescript), your command might look something like this: `node build/index.js .../quickstart-resources/weather-server-typescript/build/index.js`
751 </Note>
752
753 **The client will:**
754
755 1. Connect to the specified server
756 2. List available tools
757 3. Start an interactive chat session where you can:
758 * Enter queries
759 * See tool executions
760 * Get responses from Claude
761
762 ## How It Works
763
764 When you submit a query:
765
766 1. The client gets the list of available tools from the server
767 2. Your query is sent to Claude along with tool descriptions
768 3. Claude decides which tools (if any) to use
769 4. The client executes any requested tool calls through the server
770 5. Results are sent back to Claude
771 6. Claude provides a natural language response
772 7. The response is displayed to you
773
774 ## Best practices
775
776 1. **Error Handling**
777 * Use TypeScript's type system for better error detection
778 * Wrap tool calls in try-catch blocks
779 * Provide meaningful error messages
780 * Gracefully handle connection issues
781
782 2. **Security**
783 * Store API keys securely in `.env`
784 * Validate server responses
785 * Be cautious with tool permissions
786
787 ## Troubleshooting
788
789 ### Server Path Issues
790
791 * Double-check the path to your server script is correct
792 * Use the absolute path if the relative path isn't working
793 * For Windows users, make sure to use forward slashes (/) or escaped backslashes (\\) in the path
794 * Verify the server file has the correct extension (.js for Node.js or .py for Python)
795
796 Example of correct path usage:
797
798 ```bash theme={null}
799 # Relative path
800 node build/index.js ./server/build/index.js
801
802 # Absolute path
803 node build/index.js /Users/username/projects/mcp-server/build/index.js
804
805 # Windows path (either format works)
806 node build/index.js C:/projects/mcp-server/build/index.js
807 node build/index.js C:\\projects\\mcp-server\\build\\index.js
808 ```
809
810 ### Response Timing
811
812 * The first response might take up to 30 seconds to return
813 * This is normal and happens while:
814 * The server initializes
815 * Claude processes the query
816 * Tools are being executed
817 * Subsequent responses are typically faster
818 * Don't interrupt the process during this initial waiting period
819
820 ### Common Error Messages
821
822 If you see:
823
824 * `Error: Cannot find module`: Check your build folder and ensure TypeScript compilation succeeded
825 * `Connection refused`: Ensure the server is running and the path is correct
826 * `Tool execution failed`: Verify the tool's required environment variables are set
827 * `ANTHROPIC_API_KEY is not set`: Check your .env file and environment variables
828 * `TypeError`: Ensure you're using the correct types for tool arguments
829 * `BadRequestError`: Ensure you have enough credits to access the Anthropic API
830 </Tab>
831
832 <Tab title="Java">
833 <Note>
834 This is a quickstart demo based on Spring AI MCP auto-configuration and boot starters.
835 To learn how to create sync and async MCP Clients manually, consult the [Java SDK Client](https://java.sdk.modelcontextprotocol.io/) documentation
836 </Note>
837
838 This example demonstrates how to build an interactive chatbot that combines Spring AI's Model Context Protocol (MCP) with the [Brave Search MCP Server](https://github.com/modelcontextprotocol/servers-archived/tree/main/src/brave-search). The application creates a conversational interface powered by Anthropic's Claude AI model that can perform internet searches through Brave Search, enabling natural language interactions with real-time web data.
839 [You can find the complete code for this tutorial here.](https://github.com/spring-projects/spring-ai-examples/tree/main/model-context-protocol/web-search/brave-chatbot)
840
841 ## System Requirements
842
843 Before starting, ensure your system meets these requirements:
844
845 * Java 17 or higher
846 * Maven 3.6+
847 * npx package manager
848 * Anthropic API key (Claude)
849 * Brave Search API key
850
851 ## Setting Up Your Environment
852
853 1. Install npx (Node Package eXecute):
854 First, make sure to install [npm](https://docs.npmjs.com/downloading-and-installing-node-js-and-npm)
855 and then run:
856
857 ```bash theme={null}
858 npm install -g npx
859 ```
860
861 2. Clone the repository:
862
863 ```bash theme={null}
864 git clone https://github.com/spring-projects/spring-ai-examples.git
865 cd model-context-protocol/web-search/brave-chatbot
866 ```
867
868 3. Set up your API keys:
869
870 ```bash theme={null}
871 export ANTHROPIC_API_KEY='your-anthropic-api-key-here'
872 export BRAVE_API_KEY='your-brave-api-key-here'
873 ```
874
875 4. Build the application:
876
877 ```bash theme={null}
878 ./mvnw clean install
879 ```
880
881 5. Run the application using Maven:
882 ```bash theme={null}
883 ./mvnw spring-boot:run
884 ```
885
886 <Warning>
887 Make sure you keep your `ANTHROPIC_API_KEY` and `BRAVE_API_KEY` keys secure!
888 </Warning>
889
890 ## How it Works
891
892 The application integrates Spring AI with the Brave Search MCP server through several components:
893
894 ### MCP Client Configuration
895
896 1. Required dependencies in pom.xml:
897
898 ```xml theme={null}
899 <dependency>
900 <groupId>org.springframework.ai</groupId>
901 <artifactId>spring-ai-starter-mcp-client</artifactId>
902 </dependency>
903 <dependency>
904 <groupId>org.springframework.ai</groupId>
905 <artifactId>spring-ai-starter-model-anthropic</artifactId>
906 </dependency>
907 ```
908
909 2. Application properties (application.yml):
910
911 ```yml theme={null}
912 spring:
913 ai:
914 mcp:
915 client:
916 enabled: true
917 name: brave-search-client
918 version: 1.0.0
919 type: SYNC
920 request-timeout: 20s
921 stdio:
922 root-change-notification: true
923 servers-configuration: classpath:/mcp-servers-config.json
924 toolcallback:
925 enabled: true
926 anthropic:
927 api-key: ${ANTHROPIC_API_KEY}
928 ```
929
930 This activates the `spring-ai-starter-mcp-client` to create one or more `McpClient`s based on the provided server configuration.
931 The `spring.ai.mcp.client.toolcallback.enabled=true` property enables the tool callback mechanism, that automatically registers all MCP tool as spring ai tools.
932 It is disabled by default.
933
934 3. MCP Server Configuration (`mcp-servers-config.json`):
935
936 ```json theme={null}
937 {
938 "mcpServers": {
939 "brave-search": {
940 "command": "npx",
941 "args": ["-y", "@modelcontextprotocol/server-brave-search"],
942 "env": {
943 "BRAVE_API_KEY": "<PUT YOUR BRAVE API KEY>"
944 }
945 }
946 }
947 }
948 ```
949
950 ### Chat Implementation
951
952 The chatbot is implemented using Spring AI's ChatClient with MCP tool integration:
953
954 ```java theme={null}
955 var chatClient = chatClientBuilder
956 .defaultSystem("You are useful assistant, expert in AI and Java.")
957 .defaultToolCallbacks((Object[]) mcpToolAdapter.toolCallbacks())
958 .defaultAdvisors(new MessageChatMemoryAdvisor(new InMemoryChatMemory()))
959 .build();
960 ```
961
962 Key features:
963
964 * Uses Claude AI model for natural language understanding
965 * Integrates Brave Search through MCP for real-time web search capabilities
966 * Maintains conversation memory using InMemoryChatMemory
967 * Runs as an interactive command-line application
968
969 ### Build and run
970
971 ```bash theme={null}
972 ./mvnw clean install
973 java -jar ./target/ai-mcp-brave-chatbot-0.0.1-SNAPSHOT.jar
974 ```
975
976 or
977
978 ```bash theme={null}
979 ./mvnw spring-boot:run
980 ```
981
982 The application will start an interactive chat session where you can ask questions. The chatbot will use Brave Search when it needs to find information from the internet to answer your queries.
983
984 The chatbot can:
985
986 * Answer questions using its built-in knowledge
987 * Perform web searches when needed using Brave Search
988 * Remember context from previous messages in the conversation
989 * Combine information from multiple sources to provide comprehensive answers
990
991 ### Advanced Configuration
992
993 The MCP client supports additional configuration options:
994
995 * Client customization through `McpSyncClientCustomizer` or `McpAsyncClientCustomizer`
996 * Multiple clients with multiple transport types: `STDIO` and `SSE` (Server-Sent Events)
997 * Integration with Spring AI's tool execution framework
998 * Automatic client initialization and lifecycle management
999
1000 For WebFlux-based applications, you can use the WebFlux starter instead:
1001
1002 ```xml theme={null}
1003 <dependency>
1004 <groupId>org.springframework.ai</groupId>
1005 <artifactId>spring-ai-mcp-client-webflux-spring-boot-starter</artifactId>
1006 </dependency>
1007 ```
1008
1009 This provides similar functionality but uses a WebFlux-based SSE transport implementation, recommended for production deployments.
1010 </Tab>
1011
1012 <Tab title="Kotlin">
1013 [You can find the complete code for this tutorial here.](https://github.com/modelcontextprotocol/kotlin-sdk/tree/main/samples/kotlin-mcp-client)
1014
1015 ## System Requirements
1016
1017 Before starting, ensure your system meets these requirements:
1018
1019 * JDK 11 or higher
1020 * Anthropic API key (Claude)
1021
1022 ## Setting up your environment
1023
1024 First, let's install `java` and `gradle` if you haven't already.
1025 You can download `java` from [official Oracle JDK website](https://www.oracle.com/java/technologies/downloads/).
1026 Verify your `java` installation:
1027
1028 ```bash theme={null}
1029 java --version
1030 ```
1031
1032 Now, let's create and set up your project:
1033
1034 <CodeGroup>
1035 ```bash macOS/Linux theme={null}
1036 # Create a new directory for our project
1037 mkdir kotlin-mcp-client
1038 cd kotlin-mcp-client
1039
1040 # Initialize a new kotlin project
1041 gradle init
1042 ```
1043
1044 ```powershell Windows theme={null}
1045 # Create a new directory for our project
1046 md kotlin-mcp-client
1047 cd kotlin-mcp-client
1048 # Initialize a new kotlin project
1049 gradle init
1050 ```
1051 </CodeGroup>
1052
1053 After running `gradle init`, select **Application** as the project type, **Kotlin** as the programming language.
1054
1055 Alternatively, you can create a Kotlin application using the [IntelliJ IDEA project wizard](https://kotlinlang.org/docs/jvm-get-started.html).
1056
1057 After creating the project, replace the contents of your `build.gradle.kts` with:
1058
1059 ```kotlin build.gradle.kts theme={null}
1060 // Check latest versions at https://github.com/modelcontextprotocol/kotlin-sdk/releases
1061 val mcpVersion = "0.9.0"
1062 val ktorVersion = "3.2.3"
1063 val anthropicVersion = "2.15.0"
1064 val slf4jVersion = "2.0.17"
1065
1066 plugins {
1067 kotlin("jvm") version "2.3.20"
1068 id("com.gradleup.shadow") version "8.3.9"
1069 application
1070 }
1071
1072 application {
1073 mainClass.set("MainKt")
1074 }
1075
1076 dependencies {
1077 implementation("io.modelcontextprotocol:kotlin-sdk:$mcpVersion")
1078 implementation("io.ktor:ktor-client-cio:$ktorVersion")
1079 implementation("com.anthropic:anthropic-java:$anthropicVersion")
1080 implementation("org.slf4j:slf4j-simple:$slf4jVersion")
1081 }
1082 ```
1083
1084 Verify that everything is set up correctly:
1085
1086 ```bash theme={null}
1087 ./gradlew build
1088 ```
1089
1090 ## Setting up your API key
1091
1092 You'll need an Anthropic API key from the [Anthropic Console](https://console.anthropic.com/settings/keys).
1093
1094 Set up your API key:
1095
1096 ```bash theme={null}
1097 export ANTHROPIC_API_KEY='your-anthropic-api-key-here'
1098 ```
1099
1100 <Warning>
1101 Make sure you keep your `ANTHROPIC_API_KEY` secure!
1102 </Warning>
1103
1104 ## Creating the Client
1105
1106 ### Basic Client Structure
1107
1108 First, let's create the basic client class:
1109
1110 ```kotlin theme={null}
1111 class MCPClient(apiKey: String) : AutoCloseable {
1112 private val anthropic = AnthropicOkHttpClient.builder()
1113 .apiKey(apiKey)
1114 .build()
1115
1116 private val mcp: Client = Client(
1117 clientInfo = Implementation(name = "mcp-client-cli", version = "1.0.0")
1118 )
1119 private var serverProcess: Process? = null
1120 private lateinit var tools: List<ToolUnion>
1121
1122 // methods will go here
1123
1124 override fun close() {
1125 runBlocking {
1126 mcp.close()
1127 }
1128 serverProcess?.destroy()
1129 anthropic.close()
1130 }
1131 }
1132 ```
1133
1134 ### Server connection management
1135
1136 Next, we'll implement the method to connect to an MCP server:
1137
1138 ```kotlin theme={null}
1139 suspend fun connectToServer(serverScriptPath: String) {
1140 val command = buildList {
1141 when (serverScriptPath.substringAfterLast(".")) {
1142 "js" -> add("node")
1143 "py" -> add(if (System.getProperty("os.name").lowercase().contains("win")) "python" else "python3")
1144 "jar" -> addAll(listOf("java", "-jar"))
1145 else -> throw IllegalArgumentException("Server script must be a .js, .py or .jar file")
1146 }
1147 add(serverScriptPath)
1148 }
1149
1150 val process = ProcessBuilder(command).start()
1151 serverProcess = process
1152
1153 val transport = StdioClientTransport(
1154 input = process.inputStream.asSource().buffered(),
1155 output = process.outputStream.asSink().buffered(),
1156 )
1157
1158 mcp.connect(transport)
1159
1160 val toolsResult = mcp.listTools()
1161 tools = toolsResult.tools.map { tool ->
1162 ToolUnion.ofTool(
1163 Tool.builder()
1164 .name(tool.name)
1165 .description(tool.description ?: "")
1166 .inputSchema(
1167 Tool.InputSchema.builder()
1168 .type(JsonValue.from(tool.inputSchema.type))
1169 .properties(tool.inputSchema.properties?.toJsonValue() ?: EmptyJsonObject.toJsonValue())
1170 .putAdditionalProperty("required", JsonValue.from(tool.inputSchema.required))
1171 .build(),
1172 )
1173 .build(),
1174 )
1175 }
1176 println("Connected to server with tools: ${tools.joinToString(", ") { it.tool().get().name() }}")
1177 }
1178 ```
1179
1180 <Accordion title="JsonObject.toJsonValue() helper">
1181 This helper converts a kotlinx.serialization `JsonObject` to an Anthropic SDK `JsonValue` using Jackson:
1182
1183 ```kotlin theme={null}
1184 private fun JsonObject.toJsonValue(): JsonValue {
1185 val mapper = ObjectMapper()
1186 val node = mapper.readTree(this.toString())
1187 return JsonValue.fromJsonNode(node)
1188 }
1189 ```
1190 </Accordion>
1191
1192 ### Query processing logic
1193
1194 Now let's add the core functionality for processing queries and handling tool calls:
1195
1196 ```kotlin theme={null}
1197 suspend fun processQuery(query: String): String {
1198 val messages = mutableListOf(
1199 MessageParam.builder()
1200 .role(MessageParam.Role.USER)
1201 .content(query)
1142 val process = ProcessBuilder(command).start()
1143 serverProcess = process
1144
1145 val transport = StdioClientTransport(
1146 input = process.inputStream.asSource().buffered(),
1147 output = process.outputStream.asSink().buffered(),
1148 )
1149
1150 mcp.connect(transport)
1151
1152 val toolsResult = mcp.listTools()
1153 tools = toolsResult.tools.map { tool ->
1154 ToolUnion.ofTool(
1155 Tool.builder()
1156 .name(tool.name)
1157 .description(tool.description ?: "")
1158 .inputSchema(
1159 Tool.InputSchema.builder()
1160 .type(JsonValue.from(tool.inputSchema.type))
1161 .properties(tool.inputSchema.properties?.toJsonValue() ?: EmptyJsonObject.toJsonValue())
1162 .putAdditionalProperty("required", JsonValue.from(tool.inputSchema.required))
1163 .build(),
1164 )
12021165 .build(),
12031166 )
1204
1205 val response = anthropic.messages().create(
1206 MessageCreateParams.builder()
1207 .model("claude-sonnet-4-20250514")
1208 .maxTokens(1024)
1209 .messages(messages)
1210 .tools(tools)
1211 .build(),
1212 )
1213
1214 val finalText = mutableListOf<String>()
1215 response.content().forEach { content ->
1216 when {
1217 content.isText() -> finalText.add(content.text().get().text())
1218
1219 content.isToolUse() -> {
1220 val toolName = content.toolUse().get().name()
1221 val toolArgs =
1222 content.toolUse().get()._input().convert(object : TypeReference<Map<String, JsonValue>>() {})
1223
1224 val result = mcp.callTool(
1225 name = toolName,
1226 arguments = toolArgs ?: emptyMap(),
1227 )
1228 finalText.add("[Calling tool $toolName with args $toolArgs]")
1229
1230 messages.add(
1231 MessageParam.builder()
1232 .role(MessageParam.Role.USER)
1233 .content(
1234 result.content
1235 .filterIsInstance<TextContent>()
1236 .joinToString("\n") { it.text }
1237 )
1238 .build(),
1239 )
1240
1241 val aiResponse = anthropic.messages().create(
1242 MessageCreateParams.builder()
1243 .model("claude-sonnet-4-20250514")
1244 .maxTokens(1024)
1245 .messages(messages)
1246 .build(),
1247 )
1248
1249 finalText.add(aiResponse.content().first().text().get().text())
1250 }
1251 }
1252 }
1253
1254 return finalText.joinToString("\n")
12551167 }
1256 ```
1257
1258 ### Interactive chat
1259
1260 We'll add the chat loop:
1261
1262 ```kotlin theme={null}
1263 suspend fun chatLoop() {
1264 println("\nMCP Client Started!")
1265 println("Type your queries or 'quit' to exit.")
1266
1267 while (true) {
1268 print("\nQuery: ")
1269 val message = readlnOrNull() ?: break
1270 if (message.trim().lowercase() == "quit") break
1271
1272 try {
1273 val response = processQuery(message)
1274 println("\n$response")
1275 } catch (e: Exception) {
1276 println("\nError: ${e.message}")
1168 println("Connected to server with tools: ${tools.joinToString(", ") { it.tool().get().name() }}")
1169}
1170```
1171
1172> **JsonObject.toJsonValue() helper**
1173>
1174> This helper converts a kotlinx.serialization `JsonObject` to an Anthropic SDK `JsonValue` using Jackson:
1175>
1176> ```kotlin theme={null}
1177> private fun JsonObject.toJsonValue(): JsonValue {
1178> val mapper = ObjectMapper()
1179> val node = mapper.readTree(this.toString())
1180> return JsonValue.fromJsonNode(node)
1181> }
1182> ```
1183
1184### Query processing logic
1185
1186Now let's add the core functionality for processing queries and handling tool calls:
1187
1188```kotlin theme={null}
1189suspend fun processQuery(query: String): String {
1190 val messages = mutableListOf(
1191 MessageParam.builder()
1192 .role(MessageParam.Role.USER)
1193 .content(query)
1194 .build(),
1195 )
1196
1197 val response = anthropic.messages().create(
1198 MessageCreateParams.builder()
1199 .model("claude-sonnet-4-20250514")
1200 .maxTokens(1024)
1201 .messages(messages)
1202 .tools(tools)
1203 .build(),
1204 )
1205
1206 val finalText = mutableListOf<String>()
1207 response.content().forEach { content ->
1208 when {
1209 content.isText() -> finalText.add(content.text().get().text())
1210
1211 content.isToolUse() -> {
1212 val toolName = content.toolUse().get().name()
1213 val toolArgs =
1214 content.toolUse().get()._input().convert(object : TypeReference<Map<String, JsonValue>>() {})
1215
1216 val result = mcp.callTool(
1217 name = toolName,
1218 arguments = toolArgs ?: emptyMap(),
1219 )
1220 finalText.add("[Calling tool $toolName with args $toolArgs]")
1221
1222 messages.add(
1223 MessageParam.builder()
1224 .role(MessageParam.Role.USER)
1225 .content(
1226 result.content
1227 .filterIsInstance<TextContent>()
1228 .joinToString("\n") { it.text }
1229 )
1230 .build(),
1231 )
1232
1233 val aiResponse = anthropic.messages().create(
1234 MessageCreateParams.builder()
1235 .model("claude-sonnet-4-20250514")
1236 .maxTokens(1024)
1237 .messages(messages)
1238 .build(),
1239 )
1240
1241 finalText.add(aiResponse.content().first().text().get().text())
12771242 }
12781243 }
12791244 }
1280 ```
1281
1282 ### Main entry point
1283
1284 Finally, we'll add the main execution function:
1285
1286 ```kotlin theme={null}
1287 fun main(args: Array<String>) = runBlocking {
1288 require(args.isNotEmpty()) { "Usage: java -jar <path> <path_to_server_script>" }
1289
1290 val apiKey = System.getenv("ANTHROPIC_API_KEY")
1291 require(!apiKey.isNullOrBlank()) { "ANTHROPIC_API_KEY environment variable is not set" }
1292
1293 val client = MCPClient(apiKey)
1294 client.use {
1295 client.connectToServer(args.first())
1296 client.chatLoop()
1245
1246 return finalText.joinToString("\n")
1247}
1248```
1249
1250### Interactive chat
1251
1252We'll add the chat loop:
1253
1254```kotlin theme={null}
1255suspend fun chatLoop() {
1256 println("\nMCP Client Started!")
1257 println("Type your queries or 'quit' to exit.")
1258
1259 while (true) {
1260 print("\nQuery: ")
1261 val message = readlnOrNull() ?: break
1262 if (message.trim().lowercase() == "quit") break
1263
1264 try {
1265 val response = processQuery(message)
1266 println("\n$response")
1267 } catch (e: Exception) {
1268 println("\nError: ${e.message}")
12971269 }
12981270 }
1299 ```
1300
1301 ## Running the client
1302
1303 To run your client with any MCP server:
1304
1305 ```bash theme={null}
1306 ./gradlew build
1307
1308 # Run the client
1309 java -jar build/libs/kotlin-mcp-client-0.1.0-all.jar path/to/server.jar # JVM server
1310 java -jar build/libs/kotlin-mcp-client-0.1.0-all.jar path/to/server.py # Python server
1311 java -jar build/libs/kotlin-mcp-client-0.1.0-all.jar path/to/build/index.js # Node server
1312 ```
1313
1314 Alternatively, you can run directly with Gradle:
1315
1316 ```bash theme={null}
1317 ./gradlew run --args="path/to/server.jar"
1318 ```
1319
1320 <Note>
1321 If you're continuing the weather tutorial from the server quickstart, your command might look something like this: `java -jar build/libs/kotlin-mcp-client-0.1.0-all.jar .../samples/weather-stdio-server/build/libs/weather-stdio-server-0.1.0-all.jar`
1322 </Note>
1323
1324 **The client will:**
1325
1326 1. Connect to the specified server
1327 2. List available tools
1328 3. Start an interactive chat session where you can:
1329 * Enter queries
1330 * See tool executions
1331 * Get responses from Claude
1332
1333 ## How it works
1334
1335 Here's a high-level workflow schema:
1336
1337 ```mermaid theme={null}
1338 ---
1339 config:
1340 theme: neutral
1341 ---
1342 sequenceDiagram
1343 actor User
1344 participant Client
1345 participant Claude
1346 participant MCP_Server as MCP Server
1347 participant Tools
1348
1349 User->>Client: Send query
1350 Client<<->>MCP_Server: Get available tools
1351 Client->>Claude: Send query with tool descriptions
1352 Claude-->>Client: Decide tool execution
1353 Client->>MCP_Server: Request tool execution
1354 MCP_Server->>Tools: Execute chosen tools
1355 Tools-->>MCP_Server: Return results
1356 MCP_Server-->>Client: Send results
1357 Client->>Claude: Send tool results
1358 Claude-->>Client: Provide final response
1359 Client-->>User: Display response
1360 ```
1361
1362 When you submit a query:
1363
1364 1. The client gets the list of available tools from the server
1365 2. Your query is sent to Claude along with tool descriptions
1366 3. Claude decides which tools (if any) to use
1367 4. The client executes any requested tool calls through the server
1368 5. Results are sent back to Claude
1369 6. Claude provides a natural language response
1370 7. The response is displayed to you
1371
1372 ## Best practices
1373
1374 1. **Error Handling**
1375 * Leverage Kotlin's type system to model errors explicitly
1376 * Wrap external tool and API calls in `try-catch` blocks when exceptions are possible
1377 * Provide clear and meaningful error messages
1378 * Handle network timeouts and connection issues gracefully
1379
1380 2. **Security**
1381 * Store API keys and secrets securely in `local.properties`, environment variables, or secret managers
1382 * Validate all external responses to avoid unexpected or unsafe data usage
1383 * Be cautious with permissions and trust boundaries when using tools
1384
1385 3. **Environment**
1386 * Set `ANTHROPIC_API_KEY` through environment variables rather than hardcoding
1387 * Use `.env` files with appropriate `.gitignore` rules for local development
1388
1389 ## Troubleshooting
1390
1391 ### Server Path Issues
1392
1393 * Double-check the path to your server script is correct
1394 * Use the absolute path if the relative path isn't working
1395 * For Windows users, make sure to use forward slashes (/) or escaped backslashes (\\) in the path
1396 * Make sure that the required runtime is installed (java for Java, npm for Node.js, or uv for Python)
1397 * Verify the server file has the correct extension (.jar for Java, .js for Node.js or .py for Python)
1398
1399 Example of correct path usage:
1400
1401 ```bash theme={null}
1402 # Relative path
1403 java -jar build/libs/client.jar ./server/build/libs/server.jar
1404
1405 # Absolute path
1406 java -jar build/libs/client.jar /Users/username/projects/mcp-server/build/libs/server.jar
1407
1408 # Windows path (either format works)
1409 java -jar build/libs/client.jar C:/projects/mcp-server/build/libs/server.jar
1410 java -jar build/libs/client.jar C:\\projects\\mcp-server\\build\\libs\\server.jar
1411 ```
1412
1413 ### Build Issues
1414
1415 * Use `./gradlew build` or `./gradlew shadowJar` (not `./gradlew jar`) to create the shadow JAR with all dependencies
1416 * If you get JDK version errors, ensure your installed JDK version matches or exceeds the `jvmToolchain` setting in `build.gradle.kts`
1417
1418 ### Response Timing
1419
1420 * The first response might take up to 30 seconds to return
1421 * This is normal and happens while:
1422 * The server initializes
1423 * Claude processes the query
1424 * Tools are being executed
1425 * Subsequent responses are typically faster
1426 * Don't interrupt the process during this initial waiting period
1427
1428 ### Common Error Messages
1429
1430 If you see:
1431
1432 * `Connection refused`: Ensure the server is running and the path is correct
1433 * `Tool execution failed`: Verify the tool's required environment variables are set
1434 * `ANTHROPIC_API_KEY is not set`: Check your environment variables
1435 </Tab>
1436
1437 <Tab title="C#">
1438 [You can find the complete code for this tutorial here.](https://github.com/modelcontextprotocol/csharp-sdk/tree/main/samples/QuickstartClient)
1439
1440 ## System Requirements
1441
1442 Before starting, ensure your system meets these requirements:
1443
1444 * .NET 8.0 or higher
1445 * Anthropic API key (Claude)
1446 * Windows, Linux, or macOS
1447
1448 ## Setting up your environment
1449
1450 First, create a new .NET project:
1451
1452 ```bash theme={null}
1453 dotnet new console -n QuickstartClient
1454 cd QuickstartClient
1455 ```
1456
1457 Then, add the required dependencies to your project:
1458
1459 ```bash theme={null}
1460 dotnet add package ModelContextProtocol --prerelease
1461 dotnet add package Anthropic.SDK
1462 dotnet add package Microsoft.Extensions.Hosting
1463 dotnet add package Microsoft.Extensions.AI
1464 ```
1465
1466 ## Setting up your API key
1467
1468 You'll need an Anthropic API key from the [Anthropic Console](https://console.anthropic.com/settings/keys).
1469
1470 ```bash theme={null}
1471 dotnet user-secrets init
1472 dotnet user-secrets set "ANTHROPIC_API_KEY" "<your key here>"
1473 ```
1474
1475 ## Creating the Client
1476
1477 ### Basic Client Structure
1478
1479 First, let's setup the basic client class in the file `Program.cs`:
1480
1481 ```csharp theme={null}
1482 using Anthropic.SDK;
1483 using Microsoft.Extensions.AI;
1484 using Microsoft.Extensions.Configuration;
1485 using Microsoft.Extensions.Hosting;
1486 using ModelContextProtocol.Client;
1487 using ModelContextProtocol.Protocol.Transport;
1488
1489 var builder = Host.CreateApplicationBuilder(args);
1490
1491 builder.Configuration
1492 .AddEnvironmentVariables()
1493 .AddUserSecrets<Program>();
1494 ```
1495
1496 This creates the beginnings of a .NET console application that can read the API key from user secrets.
1497
1498 Next, we'll setup the MCP Client:
1499
1500 ```csharp theme={null}
1501 var (command, arguments) = GetCommandAndArguments(args);
1502
1503 var clientTransport = new StdioClientTransport(new()
1271}
1272```
1273
1274### Main entry point
1275
1276Finally, we'll add the main execution function:
1277
1278```kotlin theme={null}
1279fun main(args: Array<String>) = runBlocking {
1280 require(args.isNotEmpty()) { "Usage: java -jar <path> <path_to_server_script>" }
1281
1282 val apiKey = System.getenv("ANTHROPIC_API_KEY")
1283 require(!apiKey.isNullOrBlank()) { "ANTHROPIC_API_KEY environment variable is not set" }
1284
1285 val client = MCPClient(apiKey)
1286 client.use {
1287 client.connectToServer(args.first())
1288 client.chatLoop()
1289 }
1290}
1291```
1292
1293## Running the client
1294
1295To run your client with any MCP server:
1296
1297```bash theme={null}
1298./gradlew build
1299
1300# Run the client
1301java -jar build/libs/kotlin-mcp-client-0.1.0-all.jar path/to/server.jar # JVM server
1302java -jar build/libs/kotlin-mcp-client-0.1.0-all.jar path/to/server.py # Python server
1303java -jar build/libs/kotlin-mcp-client-0.1.0-all.jar path/to/build/index.js # Node server
1304```
1305
1306Alternatively, you can run directly with Gradle:
1307
1308```bash theme={null}
1309./gradlew run --args="path/to/server.jar"
1310```
1311
1312> [!NOTE]
1313>
1314> If you're continuing the weather tutorial from the server quickstart, your command might look something like this: `java -jar build/libs/kotlin-mcp-client-0.1.0-all.jar .../samples/weather-stdio-server/build/libs/weather-stdio-server-0.1.0-all.jar`
1315
1316**The client will:**
1317
13181. Connect to the specified server
13192. List available tools
13203. Start an interactive chat session where you can:
1321 * Enter queries
1322 * See tool executions
1323 * Get responses from Claude
1324
1325## How it works
1326
1327Here's a high-level workflow schema:
1328
1329```mermaid theme={null}
1330---
1331config:
1332 theme: neutral
1333---
1334sequenceDiagram
1335 actor User
1336 participant Client
1337 participant Claude
1338 participant MCP_Server as MCP Server
1339 participant Tools
1340
1341 User->>Client: Send query
1342 Client<<->>MCP_Server: Get available tools
1343 Client->>Claude: Send query with tool descriptions
1344 Claude-->>Client: Decide tool execution
1345 Client->>MCP_Server: Request tool execution
1346 MCP_Server->>Tools: Execute chosen tools
1347 Tools-->>MCP_Server: Return results
1348 MCP_Server-->>Client: Send results
1349 Client->>Claude: Send tool results
1350 Claude-->>Client: Provide final response
1351 Client-->>User: Display response
1352```
1353
1354When you submit a query:
1355
13561. The client gets the list of available tools from the server
13572. Your query is sent to Claude along with tool descriptions
13583. Claude decides which tools (if any) to use
13594. The client executes any requested tool calls through the server
13605. Results are sent back to Claude
13616. Claude provides a natural language response
13627. The response is displayed to you
1363
1364## Best practices
1365
13661. **Error Handling**
1367 * Leverage Kotlin's type system to model errors explicitly
1368 * Wrap external tool and API calls in `try-catch` blocks when exceptions are possible
1369 * Provide clear and meaningful error messages
1370 * Handle network timeouts and connection issues gracefully
1371
13722. **Security**
1373 * Store API keys and secrets securely in `local.properties`, environment variables, or secret managers
1374 * Validate all external responses to avoid unexpected or unsafe data usage
1375 * Be cautious with permissions and trust boundaries when using tools
1376
13773. **Environment**
1378 * Set `ANTHROPIC_API_KEY` through environment variables rather than hardcoding
1379 * Use `.env` files with appropriate `.gitignore` rules for local development
1380
1381## Troubleshooting
1382
1383### Server Path Issues
1384
1385* Double-check the path to your server script is correct
1386* Use the absolute path if the relative path isn't working
1387* For Windows users, make sure to use forward slashes (/) or escaped backslashes (\\) in the path
1388* Make sure that the required runtime is installed (java for Java, npm for Node.js, or uv for Python)
1389* Verify the server file has the correct extension (.jar for Java, .js for Node.js or .py for Python)
1390
1391Example of correct path usage:
1392
1393```bash theme={null}
1394# Relative path
1395java -jar build/libs/client.jar ./server/build/libs/server.jar
1396
1397# Absolute path
1398java -jar build/libs/client.jar /Users/username/projects/mcp-server/build/libs/server.jar
1399
1400# Windows path (either format works)
1401java -jar build/libs/client.jar C:/projects/mcp-server/build/libs/server.jar
1402java -jar build/libs/client.jar C:\\projects\\mcp-server\\build\\libs\\server.jar
1403```
1404
1405### Build Issues
1406
1407* Use `./gradlew build` or `./gradlew shadowJar` (not `./gradlew jar`) to create the shadow JAR with all dependencies
1408* If you get JDK version errors, ensure your installed JDK version matches or exceeds the `jvmToolchain` setting in `build.gradle.kts`
1409
1410### Response Timing
1411
1412* The first response might take up to 30 seconds to return
1413* This is normal and happens while:
1414 * The server initializes
1415 * Claude processes the query
1416 * Tools are being executed
1417* Subsequent responses are typically faster
1418* Don't interrupt the process during this initial waiting period
1419
1420### Common Error Messages
1421
1422If you see:
1423
1424* `Connection refused`: Ensure the server is running and the path is correct
1425* `Tool execution failed`: Verify the tool's required environment variables are set
1426* `ANTHROPIC_API_KEY is not set`: Check your environment variables
1427
1428**C#**
1429
1430[You can find the complete code for this tutorial here.](https://github.com/modelcontextprotocol/csharp-sdk/tree/main/samples/QuickstartClient)
1431
1432## System Requirements
1433
1434Before starting, ensure your system meets these requirements:
1435
1436* .NET 8.0 or higher
1437* Anthropic API key (Claude)
1438* Windows, Linux, or macOS
1439
1440## Setting up your environment
1441
1442First, create a new .NET project:
1443
1444```bash theme={null}
1445dotnet new console -n QuickstartClient
1446cd QuickstartClient
1447```
1448
1449Then, add the required dependencies to your project:
1450
1451```bash theme={null}
1452dotnet add package ModelContextProtocol --prerelease
1453dotnet add package Anthropic.SDK
1454dotnet add package Microsoft.Extensions.Hosting
1455dotnet add package Microsoft.Extensions.AI
1456```
1457
1458## Setting up your API key
1459
1460You'll need an Anthropic API key from the [Anthropic Console](https://console.anthropic.com/settings/keys).
1461
1462```bash theme={null}
1463dotnet user-secrets init
1464dotnet user-secrets set "ANTHROPIC_API_KEY" "<your key here>"
1465```
1466
1467## Creating the Client
1468
1469### Basic Client Structure
1470
1471First, let's setup the basic client class in the file `Program.cs`:
1472
1473```csharp theme={null}
1474using Anthropic.SDK;
1475using Microsoft.Extensions.AI;
1476using Microsoft.Extensions.Configuration;
1477using Microsoft.Extensions.Hosting;
1478using ModelContextProtocol.Client;
1479using ModelContextProtocol.Protocol.Transport;
1480
1481var builder = Host.CreateApplicationBuilder(args);
1482
1483builder.Configuration
1484 .AddEnvironmentVariables()
1485 .AddUserSecrets<Program>();
1486```
1487
1488This creates the beginnings of a .NET console application that can read the API key from user secrets.
1489
1490Next, we'll setup the MCP Client:
1491
1492```csharp theme={null}
1493var (command, arguments) = GetCommandAndArguments(args);
1494
1495var clientTransport = new StdioClientTransport(new()
1496{
1497 Name = "Demo Server",
1498 Command = command,
1499 Arguments = arguments,
1500});
1501
1502await using var mcpClient = await McpClient.CreateAsync(clientTransport);
1503
1504var tools = await mcpClient.ListToolsAsync();
1505foreach (var tool in tools)
1506{
1507 Console.WriteLine($"Connected to server with tools: {tool.Name}");
1508}
1509```
1510
1511Add this function at the end of the `Program.cs` file:
1512
1513```csharp theme={null}
1514static (string command, string[] arguments) GetCommandAndArguments(string[] args)
1515{
1516 return args switch
15041517 {
1505 Name = "Demo Server",
1506 Command = command,
1507 Arguments = arguments,
1508 });
1509
1510 await using var mcpClient = await McpClient.CreateAsync(clientTransport);
1511
1512 var tools = await mcpClient.ListToolsAsync();
1513 foreach (var tool in tools)
1518 [var script] when script.EndsWith(".py") => ("python", args),
1519 [var script] when script.EndsWith(".js") => ("node", args),
1520 [var script] when Directory.Exists(script) || (File.Exists(script) && script.EndsWith(".csproj")) => ("dotnet", ["run", "--project", script, "--no-build"]),
1521 _ => throw new NotSupportedException("An unsupported server script was provided. Supported scripts are .py, .js, or .csproj")
1522 };
1523}
1524```
1525
1526This creates an MCP client that will connect to a server that is provided as a command line argument. It then lists the available tools from the connected server.
1527
1528### Query processing logic
1529
1530Now let's add the core functionality for processing queries and handling tool calls:
1531
1532```csharp theme={null}
1533using var anthropicClient = new AnthropicClient(new APIAuthentication(builder.Configuration["ANTHROPIC_API_KEY"]))
1534 .Messages
1535 .AsBuilder()
1536 .UseFunctionInvocation()
1537 .Build();
1538
1539var options = new ChatOptions
1540{
1541 MaxOutputTokens = 1000,
1542 ModelId = "claude-sonnet-4-20250514",
1543 Tools = [.. tools]
1544};
1545
1546Console.ForegroundColor = ConsoleColor.Green;
1547Console.WriteLine("MCP Client Started!");
1548Console.ResetColor();
1549
1550PromptForInput();
1551while(Console.ReadLine() is string query && !"exit".Equals(query, StringComparison.OrdinalIgnoreCase))
1552{
1553 if (string.IsNullOrWhiteSpace(query))
15141554 {
1515 Console.WriteLine($"Connected to server with tools: {tool.Name}");
1555 PromptForInput();
1556 continue;
15161557 }
1517 ```
1518
1519 Add this function at the end of the `Program.cs` file:
1520
1521 ```csharp theme={null}
1522 static (string command, string[] arguments) GetCommandAndArguments(string[] args)
1558
1559 await foreach (var message in anthropicClient.GetStreamingResponseAsync(query, options))
15231560 {
1524 return args switch
1525 {
1526 [var script] when script.EndsWith(".py") => ("python", args),
1527 [var script] when script.EndsWith(".js") => ("node", args),
1528 [var script] when Directory.Exists(script) || (File.Exists(script) && script.EndsWith(".csproj")) => ("dotnet", ["run", "--project", script, "--no-build"]),
1529 _ => throw new NotSupportedException("An unsupported server script was provided. Supported scripts are .py, .js, or .csproj")
1530 };
1561 Console.Write(message);
15311562 }
1532 ```
1533
1534 This creates an MCP client that will connect to a server that is provided as a command line argument. It then lists the available tools from the connected server.
1535
1536 ### Query processing logic
1537
1538 Now let's add the core functionality for processing queries and handling tool calls:
1539
1540 ```csharp theme={null}
1541 using var anthropicClient = new AnthropicClient(new APIAuthentication(builder.Configuration["ANTHROPIC_API_KEY"]))
1542 .Messages
1543 .AsBuilder()
1544 .UseFunctionInvocation()
1545 .Build();
1546
1547 var options = new ChatOptions
1548 {
1549 MaxOutputTokens = 1000,
1550 ModelId = "claude-sonnet-4-20250514",
1551 Tools = [.. tools]
1552 };
1553
1554 Console.ForegroundColor = ConsoleColor.Green;
1555 Console.WriteLine("MCP Client Started!");
1563 Console.WriteLine();
1564
1565 PromptForInput();
1566}
1567
1568static void PromptForInput()
1569{
1570 Console.WriteLine("Enter a command (or 'exit' to quit):");
1571 Console.ForegroundColor = ConsoleColor.Cyan;
1572 Console.Write("> ");
15561573 Console.ResetColor();
1557
1558 PromptForInput();
1559 while(Console.ReadLine() is string query && !"exit".Equals(query, StringComparison.OrdinalIgnoreCase))
1560 {
1561 if (string.IsNullOrWhiteSpace(query))
1562 {
1563 PromptForInput();
1564 continue;
1565 }
1566
1567 await foreach (var message in anthropicClient.GetStreamingResponseAsync(query, options))
1568 {
1569 Console.Write(message);
1570 }
1571 Console.WriteLine();
1572
1573 PromptForInput();
1574 }
1575
1576 static void PromptForInput()
1577 {
1578 Console.WriteLine("Enter a command (or 'exit' to quit):");
1579 Console.ForegroundColor = ConsoleColor.Cyan;
1580 Console.Write("> ");
1581 Console.ResetColor();
1582 }
1583 ```
1584
1585 ## Key Components Explained
1586
1587 ### 1. Client Initialization
1588
1589 * The client is initialized using `McpClient.CreateAsync()`, which sets up the transport type and command to run the server.
1590
1591 ### 2. Server Connection
1592
1593 * Supports Python, Node.js, and .NET servers.
1594 * The server is started using the command specified in the arguments.
1595 * Configures to use stdio for communication with the server.
1596 * Initializes the session and available tools.
1597
1598 ### 3. Query Processing
1599
1600 * Leverages [Microsoft.Extensions.AI](https://learn.microsoft.com/dotnet/ai/ai-extensions) for the chat client.
1601 * Configures the `IChatClient` to use automatic tool (function) invocation.
1602 * The client reads user input and sends it to the server.
1603 * The server processes the query and returns a response.
1604 * The response is displayed to the user.
1605
1606 ## Running the Client
1607
1608 To run your client with any MCP server:
1609
1610 ```bash theme={null}
1611 dotnet run -- path/to/server.csproj # dotnet server
1612 dotnet run -- path/to/server.py # python server
1613 dotnet run -- path/to/server.js # node server
1614 ```
1615
1616 <Note>
1617 If you're continuing the weather tutorial from the server quickstart, your command might look something like this: `dotnet run -- path/to/QuickstartWeatherServer`.
1618 </Note>
1619
1620 The client will:
1621
1622 1. Connect to the specified server
1623 2. List available tools
1624 3. Start an interactive chat session where you can:
1625 * Enter queries
1626 * See tool executions
1627 * Get responses from Claude
1628 4. Exit the session when done
1629
1630 Here's an example of what it should look like if connected to the weather server quickstart:
1631
1632 <Frame>
1633 <img src="https://mintcdn.com/mcp/4ZXF1PrDkEaJvXpn/images/quickstart-dotnet-client.png?fit=max&auto=format&n=4ZXF1PrDkEaJvXpn&q=85&s=fcf28dde150d6db879402ad8150c6b23" width="1115" height="666" data-path="images/quickstart-dotnet-client.png" />
1634 </Frame>
1635 </Tab>
1636
1637 <Tab title="Ruby">
1638 [You can find the complete code for this tutorial here.](https://github.com/modelcontextprotocol/quickstart-resources/tree/main/mcp-client-ruby)
1639
1640 ## System Requirements
1641
1642 Before starting, ensure your system meets these requirements:
1643
1644 * Mac or Windows computer
1645 * Ruby 3.2.0 or higher installed (required by the [Anthropic SDK](https://github.com/anthropics/anthropic-sdk-ruby))
1646 * Anthropic API key (Claude)
1647
1648 ## Setting Up Your Environment
1649
1650 First, create a new Ruby project:
1651
1652 <CodeGroup>
1653 ```bash macOS/Linux theme={null}
1654 # Create project directory
1655 mkdir mcp-client
1656 cd mcp-client
1657
1658 # Create a Gemfile
1659 bundle init
1660
1661 # Add required dependencies
1662 bundle add anthropic base64 dotenv mcp
1663
1664 # Create our main file
1665 touch client.rb
1666 ```
1667
1668 ```powershell Windows theme={null}
1669 # Create project directory
1670 mkdir mcp-client
1671 cd mcp-client
1672
1673 # Create a Gemfile
1674 bundle init
1675
1676 # Add required dependencies
1677 bundle add anthropic base64 dotenv mcp
1678
1679 # Create our main file
1680 new-item client.rb
1681 ```
1682 </CodeGroup>
1683
1684 ## Setting Up Your API Key
1685
1686 You'll need an Anthropic API key from the [Anthropic Console](https://console.anthropic.com/settings/keys).
1687
1688 Create a `.env` file to store it:
1689
1690 ```bash theme={null}
1691 echo "ANTHROPIC_API_KEY=your-api-key-goes-here" > .env
1692 ```
1693
1694 Add `.env` to your `.gitignore`:
1695
1696 ```bash theme={null}
1697 echo ".env" >> .gitignore
1698 ```
1699
1700 <Warning>
1701 Make sure you keep your `ANTHROPIC_API_KEY` secure!
1702 </Warning>
1703
1704 ## Creating the Client
1705
1706 ### Basic Client Structure
1707
1708 First, let's set up our requires and create the basic client class:
1709
1710 ```ruby theme={null}
1711 require "anthropic"
1712 require "dotenv/load"
1713 require "json"
1714 require "mcp"
1715
1716 class MCPClient
1717 ANTHROPIC_MODEL = "claude-sonnet-4-20250514"
1718
1719 def initialize
1720 @mcp_client = nil
1721 @transport = nil
1722 @anthropic_client = nil
1723 end
1724
1725 # methods will go here
1726 end
1727 ```
1728
1729 ### Server Connection Management
1730
1731 Next, we'll implement the method to connect to an MCP server:
1732
1733 ```ruby theme={null}
1734 def connect_to_server(server_script_path)
1735 command = case File.extname(server_script_path)
1736 when ".rb"
1737 "ruby"
1738 when ".py"
1739 "python3"
1740 when ".js"
1741 "node"
1742 else
1743 raise ArgumentError, "Server script must be a .rb, .py, or .js file."
1744 end
1745
1746 @transport = MCP::Client::Stdio.new(command: command, args: [server_script_path])
1747 @mcp_client = MCP::Client.new(transport: @transport)
1748 @mcp_client.connect
1749
1750 tool_names = @mcp_client.tools.map(&:name)
1751 puts "\nConnected to server with tools: #{tool_names}"
1752 end
1753 ```
1754
1755 ### Query Processing Logic
1756
1757 Now let's add the core functionality for processing queries and handling tool calls:
1758
1759 ```ruby theme={null}
1760 private
1761
1762 def process_query(query)
1763 messages = [{ role: "user", content: query }]
1764
1765 available_tools = @mcp_client.tools.map do |tool|
1766 { name: tool.name, description: tool.description, input_schema: tool.input_schema }
1767 end
1768
1769 # Initial Claude API call.
1770 response = chat(messages, tools: available_tools)
1771
1772 # Process response and handle tool calls.
1773 if response.content.any?(Anthropic::Models::ToolUseBlock)
1774 assistant_content = response.content.filter_map do |content_block|
1775 case content_block
1776 when Anthropic::Models::TextBlock
1777 { type: "text", text: content_block.text }
1778 when Anthropic::Models::ToolUseBlock
1779 { type: "tool_use", id: content_block.id, name: content_block.name, input: content_block.input }
1780 end
1781 end
1782 messages << { role: "assistant", content: assistant_content }
1783 end
1784
1785 response.content.each_with_object([]) do |content, response_parts|
1786 case content
1787 when Anthropic::Models::TextBlock
1788 response_parts << content.text
1789 when Anthropic::Models::ToolUseBlock
1790 # Execute tool call via MCP.
1791 result = @mcp_client.call_tool(name: content.name, arguments: content.input)
1792 response_parts << "[Calling tool #{content.name} with args #{content.input.to_json}]"
1793
1794 tool_result_content = result.dig("result", "content")
1795 result_text = if tool_result_content.is_a?(Array)
1796 tool_result_content.filter_map { |content_item| content_item["text"] }.join("\n")
1797 else
1798 tool_result_content.to_s
1799 end
1800
1801 messages << {
1802 role: "user",
1803 content: [{
1804 type: "tool_result",
1805 tool_use_id: content.id,
1806 content: result_text
1807 }]
1808 }
1809
1810 # Get next response from Claude.
1811 response = chat(messages)
1812
1813 response.content.each do |content_block|
1814 response_parts << content_block.text if content_block.is_a?(Anthropic::Models::TextBlock)
1815 end
1816 end
1817 end.join("\n")
1818 end
1819
1820 def chat(messages, tools: nil)
1821 params = { model: ANTHROPIC_MODEL, max_tokens: 1000, messages: messages }
1822 params[:tools] = tools if tools
1823
1824 anthropic_client.messages.create(**params)
1825 end
1826
1827 def anthropic_client
1828 @anthropic_client ||= Anthropic::Client.new(api_key: ENV["ANTHROPIC_API_KEY"])
1829 end
1830 ```
1831
1832 ### Interactive Chat Interface
1833
1834 Now we'll add the chat loop and cleanup functionality:
1835
1836 ```ruby theme={null}
1837 def chat_loop
1838 puts <<~MESSAGE
1839 MCP Client Started!
1840 Type your queries or 'quit' to exit.
1841 MESSAGE
1842
1843 loop do
1844 print "\nQuery: "
1845 line = $stdin.gets
1846 break if line.nil?
1847
1848 query = line.chomp.strip
1849 break if query.downcase == "quit"
1850 next if query.empty?
1851
1852 begin
1853 response = process_query(query)
1854 puts "\n#{response}"
1855 rescue => e
1856 puts "\nError: #{e.message}"
1857 end
1574}
1575```
1576
1577## Key Components Explained
1578
1579### 1. Client Initialization
1580
1581* The client is initialized using `McpClient.CreateAsync()`, which sets up the transport type and command to run the server.
1582
1583### 2. Server Connection
1584
1585* Supports Python, Node.js, and .NET servers.
1586* The server is started using the command specified in the arguments.
1587* Configures to use stdio for communication with the server.
1588* Initializes the session and available tools.
1589
1590### 3. Query Processing
1591
1592* Leverages [Microsoft.Extensions.AI](https://learn.microsoft.com/dotnet/ai/ai-extensions) for the chat client.
1593* Configures the `IChatClient` to use automatic tool (function) invocation.
1594* The client reads user input and sends it to the server.
1595* The server processes the query and returns a response.
1596* The response is displayed to the user.
1597
1598## Running the Client
1599
1600To run your client with any MCP server:
1601
1602```bash theme={null}
1603dotnet run -- path/to/server.csproj # dotnet server
1604dotnet run -- path/to/server.py # python server
1605dotnet run -- path/to/server.js # node server
1606```
1607
1608> [!NOTE]
1609>
1610> If you're continuing the weather tutorial from the server quickstart, your command might look something like this: `dotnet run -- path/to/QuickstartWeatherServer`.
1611
1612The client will:
1613
16141. Connect to the specified server
16152. List available tools
16163. Start an interactive chat session where you can:
1617 * Enter queries
1618 * See tool executions
1619 * Get responses from Claude
16204. Exit the session when done
1621
1622Here's an example of what it should look like if connected to the weather server quickstart:
1623
1624
1625
1626**Ruby**
1627
1628[You can find the complete code for this tutorial here.](https://github.com/modelcontextprotocol/quickstart-resources/tree/main/mcp-client-ruby)
1629
1630## System Requirements
1631
1632Before starting, ensure your system meets these requirements:
1633
1634* Mac or Windows computer
1635* Ruby 3.2.0 or higher installed (required by the [Anthropic SDK](https://github.com/anthropics/anthropic-sdk-ruby))
1636* Anthropic API key (Claude)
1637
1638## Setting Up Your Environment
1639
1640First, create a new Ruby project:
1641
1642```bash macOS/Linux theme={null}
1643# Create project directory
1644mkdir mcp-client
1645cd mcp-client
1646
1647# Create a Gemfile
1648bundle init
1649
1650# Add required dependencies
1651bundle add anthropic base64 dotenv mcp
1652
1653# Create our main file
1654touch client.rb
1655```
1656
1657```powershell Windows theme={null}
1658# Create project directory
1659mkdir mcp-client
1660cd mcp-client
1661
1662# Create a Gemfile
1663bundle init
1664
1665# Add required dependencies
1666bundle add anthropic base64 dotenv mcp
1667
1668# Create our main file
1669new-item client.rb
1670```
1671
1672## Setting Up Your API Key
1673
1674You'll need an Anthropic API key from the [Anthropic Console](https://console.anthropic.com/settings/keys).
1675
1676Create a `.env` file to store it:
1677
1678```bash theme={null}
1679echo "ANTHROPIC_API_KEY=your-api-key-goes-here" > .env
1680```
1681
1682Add `.env` to your `.gitignore`:
1683
1684```bash theme={null}
1685echo ".env" >> .gitignore
1686```
1687
1688> [!WARNING]
1689>
1690> Make sure you keep your `ANTHROPIC_API_KEY` secure!
1691
1692## Creating the Client
1693
1694### Basic Client Structure
1695
1696First, let's set up our requires and create the basic client class:
1697
1698```ruby theme={null}
1699require "anthropic"
1700require "dotenv/load"
1701require "json"
1702require "mcp"
1703
1704class MCPClient
1705 ANTHROPIC_MODEL = "claude-sonnet-4-20250514"
1706
1707 def initialize
1708 @mcp_client = nil
1709 @transport = nil
1710 @anthropic_client = nil
1711 end
1712
1713 # methods will go here
1714end
1715```
1716
1717### Server Connection Management
1718
1719Next, we'll implement the method to connect to an MCP server:
1720
1721```ruby theme={null}
1722def connect_to_server(server_script_path)
1723 command = case File.extname(server_script_path)
1724 when ".rb"
1725 "ruby"
1726 when ".py"
1727 "python3"
1728 when ".js"
1729 "node"
1730 else
1731 raise ArgumentError, "Server script must be a .rb, .py, or .js file."
1732 end
1733
1734 @transport = MCP::Client::Stdio.new(command: command, args: [server_script_path])
1735 @mcp_client = MCP::Client.new(transport: @transport)
1736 @mcp_client.connect
1737
1738 tool_names = @mcp_client.tools.map(&:name)
1739 puts "\nConnected to server with tools: #{tool_names}"
1740end
1741```
1742
1743### Query Processing Logic
1744
1745Now let's add the core functionality for processing queries and handling tool calls:
1746
1747```ruby theme={null}
1748private
1749
1750def process_query(query)
1751 messages = [{ role: "user", content: query }]
1752
1753 available_tools = @mcp_client.tools.map do |tool|
1754 { name: tool.name, description: tool.description, input_schema: tool.input_schema }
1755 end
1756
1757 # Initial Claude API call.
1758 response = chat(messages, tools: available_tools)
1759
1760 # Process response and handle tool calls.
1761 if response.content.any?(Anthropic::Models::ToolUseBlock)
1762 assistant_content = response.content.filter_map do |content_block|
1763 case content_block
1764 when Anthropic::Models::TextBlock
1765 { type: "text", text: content_block.text }
1766 when Anthropic::Models::ToolUseBlock
1767 { type: "tool_use", id: content_block.id, name: content_block.name, input: content_block.input }
18581768 end
18591769 end
1860
1861 def cleanup
1862 @transport&.close
1770 messages << { role: "assistant", content: assistant_content }
1771 end
1772
1773 response.content.each_with_object([]) do |content, response_parts|
1774 case content
1775 when Anthropic::Models::TextBlock
1776 response_parts << content.text
1777 when Anthropic::Models::ToolUseBlock
1778 # Execute tool call via MCP.
1779 result = @mcp_client.call_tool(name: content.name, arguments: content.input)
1780 response_parts << "[Calling tool #{content.name} with args #{content.input.to_json}]"
1781
1782 tool_result_content = result.dig("result", "content")
1783 result_text = if tool_result_content.is_a?(Array)
1784 tool_result_content.filter_map { |content_item| content_item["text"] }.join("\n")
1785 else
1786 tool_result_content.to_s
1787 end
1788
1789 messages << {
1790 role: "user",
1791 content: [{
1792 type: "tool_result",
1793 tool_use_id: content.id,
1794 content: result_text
1795 }]
1796 }
1797
1798 # Get next response from Claude.
1799 response = chat(messages)
1800
1801 response.content.each do |content_block|
1802 response_parts << content_block.text if content_block.is_a?(Anthropic::Models::TextBlock)
1803 end
18631804 end
1864 ```
1865
1866 ### Main Entry Point
1867
1868 Finally, we'll add the main execution logic:
1869
1870 ```ruby theme={null}
1871 if ARGV.empty?
1872 puts "Usage: ruby client.rb <path_to_server_script>"
1873 exit 1
1805 end.join("\n")
1806end
1807
1808def chat(messages, tools: nil)
1809 params = { model: ANTHROPIC_MODEL, max_tokens: 1000, messages: messages }
1810 params[:tools] = tools if tools
1811
1812 anthropic_client.messages.create(**params)
1813end
1814
1815def anthropic_client
1816 @anthropic_client ||= Anthropic::Client.new(api_key: ENV["ANTHROPIC_API_KEY"])
1817end
1818```
1819
1820### Interactive Chat Interface
1821
1822Now we'll add the chat loop and cleanup functionality:
1823
1824```ruby theme={null}
1825def chat_loop
1826 puts <<~MESSAGE
1827 MCP Client Started!
1828 Type your queries or 'quit' to exit.
1829 MESSAGE
1830
1831 loop do
1832 print "\nQuery: "
1833 line = $stdin.gets
1834 break if line.nil?
1835
1836 query = line.chomp.strip
1837 break if query.downcase == "quit"
1838 next if query.empty?
1839
1840 begin
1841 response = process_query(query)
1842 puts "\n#{response}"
1843 rescue => e
1844 puts "\nError: #{e.message}"
18741845 end
1875
1876 client = MCPClient.new
1877
1878 begin
1879 client.connect_to_server(ARGV[0])
1880
1881 api_key = ENV["ANTHROPIC_API_KEY"]
1882 if api_key.nil? || api_key.empty?
1883 puts <<~MESSAGE
1884 No ANTHROPIC_API_KEY found. To query these tools with Claude, set your API key:
1885 export ANTHROPIC_API_KEY=your-api-key-here
1886 MESSAGE
1887 exit
1888 end
1889
1890 client.chat_loop
1891 rescue => e
1892 puts "Error: #{e.message}"
1893 exit 1
1894 ensure
1895 client.cleanup
1896 end
1897 ```
1898
1899 You can find the complete `client.rb` file [here](https://github.com/modelcontextprotocol/quickstart-resources/blob/main/mcp-client-ruby/client.rb).
1900
1901 ## Key Components Explained
1902
1903 ### 1. Client Initialization
1904
1905 * The `MCPClient` class initializes with nil references for lazy setup
1906 * The Anthropic client is lazily initialized via the `anthropic_client` method
1907 * Uses `dotenv` to load environment variables from `.env`
1908
1909 ### 2. Server Connection
1910
1911 * Supports Ruby, Python, and Node.js servers
1912 * Uses `File.extname` to determine the server script type
1913 * Uses `MCP::Client::Stdio` for stdio transport
1914 * Initializes the MCP client and lists available tools
1915
1916 ### 3. Query Processing
1917
1918 * Maps MCP tools to Anthropic tool format (`name`, `description`, `input_schema`)
1919 * Uses `Anthropic::Models::TextBlock` and `Anthropic::Models::ToolUseBlock` for pattern matching
1920 * Builds assistant content once before iterating tool calls
1921 * Executes tool calls via `@mcp_client.call_tool`
1922 * Uses `chat` helper method to wrap Anthropic API calls
1923 * Extracts tool result content with `result.dig("result", "content")`
1924 * Passes tool results back to Claude for a final response
1925
1926 ### 4. Interactive Interface
1927
1928 * Provides a simple command-line interface
1929 * Handles user input and displays responses
1930 * Skips empty queries
1931 * Includes basic error handling
1932
1933 ### 5. Resource Management
1934
1935 * Proper cleanup of the transport via `begin`...`ensure`
1936 * Top-level `rescue` for error handling
1937 * API key validation after server connection
1938
1939 ## Running the Client
1940
1941 To run your client with any MCP server:
1942
1943 ```bash theme={null}
1944 bundle exec ruby client.rb path/to/server.rb # ruby server
1945 bundle exec ruby client.rb path/to/server.py # python server
1946 bundle exec ruby client.rb path/to/build/index.js # node server
1947 ```
1948
1949 <Note>
1950 If you're continuing [the weather tutorial from the server quickstart](https://github.com/modelcontextprotocol/quickstart-resources/tree/main/weather-server-ruby), your command might look something like this: `bundle exec ruby client.rb /path/to/weather-server-ruby/weather.rb`
1951 </Note>
1952
1953 The client will:
1954
1955 1. Connect to the specified server
1956 2. List available tools
1957 3. Start an interactive chat session where you can:
1958 * Enter queries
1959 * See tool executions
1960 * Get responses from Claude
1961
1962 ## How It Works
1963
1964 When you submit a query:
1965
1966 1. The client gets the list of available tools from the server
1967 2. Your query is sent to Claude along with tool descriptions
1968 3. Claude decides which tools (if any) to use
1969 4. The client executes any requested tool calls through the server
1970 5. Results are sent back to Claude
1971 6. Claude provides a natural language response
1972 7. The response is displayed to you
1973
1974 ## Best practices
1975
1976 1. **Error Handling**
1977 * Wrap tool calls in `begin`...`rescue` blocks
1978 * Provide meaningful error messages
1979 * Gracefully handle connection issues
1980
1981 2. **Resource Management**
1982 * Always close the transport when done
1983 * Use `begin`...`ensure` for proper cleanup
1984 * Handle server disconnections
1985
1986 3. **Security**
1987 * Store API keys securely in `.env`
1988 * Validate server responses
1989 * Be cautious with tool permissions
1990
1991 ## Troubleshooting
1992
1993 ### Server Path Issues
1994
1995 * Double-check the path to your server script is correct
1996 * Use the absolute path if the relative path isn't working
1997 * For Windows users, make sure to use forward slashes (/) or escaped backslashes (\\) in the path
1998 * Verify the server file has the correct extension (.py for Python, .js for Node.js, or .rb for Ruby)
1999
2000 Example of correct path usage:
2001
2002 ```bash theme={null}
2003 # Relative path
2004 bundle exec ruby client.rb ./server/weather.rb
2005
2006 # Absolute path
2007 bundle exec ruby client.rb /Users/username/projects/mcp-server/weather.rb
2008
2009 # Windows path (either format works)
2010 bundle exec ruby client.rb C:/projects/mcp-server/weather.rb
2011 bundle exec ruby client.rb C:\\projects\\mcp-server\\weather.rb
2012 ```
2013
2014 ### Response Timing
2015
2016 * The first response might take up to 30 seconds to return
2017 * This is normal and happens while:
2018 * The server initializes
2019 * Claude processes the query
2020 * Tools are being executed
2021 * Subsequent responses are typically faster
2022 * Don't interrupt the process during this initial waiting period
2023
2024 ### Common Error Messages
2025
2026 If you see:
2027
2028 * `Errno::ENOENT`: Check your server path and ensure the command (`ruby`, `python3`, `node`) is available
2029 * `Connection refused`: Ensure the server is running and the path is correct
2030 * `Tool execution failed`: Verify the tool's required environment variables are set
2031 * `Anthropic::Errors::AuthenticationError`: Check your `.env` file has a valid `ANTHROPIC_API_KEY`
2032 </Tab>
2033
2034 <Tab title="Rust">
2035 [You can find the complete code for this tutorial here.](https://github.com/modelcontextprotocol/quickstart-resources/tree/main/mcp-client-rust)
2036
2037 ## System Requirements
2038
2039 Before starting, ensure your Linux system meets these requirements:
2040
2041 * Latest stable version of [Rust and Cargo](https://www.rust-lang.org/tools/install)
2042 * Anthropic API key (Claude)
2043 * A Python, Node.js, or executable MCP server to connect to
2044
2045 ## Setting Up Your Environment
2046
2047 First, create a new Rust project:
2048
2049 ```bash theme={null}
2050 cargo new mcp-client-rust
2051 cd mcp-client-rust
2052 ```
2053
2054 Replace the contents of `Cargo.toml` with the following:
2055
2056 ```toml Cargo.toml theme={null}
2057 [package]
2058 name = "mcp-client-rust"
2059 version = "0.1.0"
2060 edition = "2024"
2061
2062 [dependencies]
2063 anyhow = "1.0.100"
2064 genai = "0.4.2"
2065 rmcp = { version = "0.8.0", features = ["server", "client", "transport-io", "transport-child-process"] }
2066 tokio = { version = "1.47.1", features = ["full"] }
2067 tracing = "0.1.41"
2068 tracing-subscriber = { version = "0.3", features = ["env-filter"] }
2069 serde_json = "1.0.128"
2070 dotenvy = "0.15.7"
2071 reqwest = "0.12.23"
2072 ```
2073
2074 The [`rmcp`](https://github.com/modelcontextprotocol/rust-sdk) crate provides the Rust MCP SDK and child-process transport. This example uses the [`genai`](https://github.com/jeremychone/rust-genai) crate to send requests to Claude and represent tools in the model request.
2075
2076 ## Setting Up Your API Key
2077
2078 You'll need an Anthropic API key from the [Anthropic Console](https://console.anthropic.com/settings/keys).
2079
2080 Create a `.env` file to store it:
2081
2082 ```bash theme={null}
2083 echo "ANTHROPIC_API_KEY=your-api-key-goes-here" > .env
2084 ```
2085
2086 Add `.env` to your `.gitignore`:
2087
2088 ```bash theme={null}
2089 echo ".env" >> .gitignore
2090 ```
2091
2092 <Warning>
2093 Make sure you keep your `ANTHROPIC_API_KEY` secure!
2094 </Warning>
2095
2096 ## Creating the Client
2097
2098 Open `src/main.rs` and replace its contents as you work through the following sections.
2099
2100 ### Imports and Client Structure
2101
2102 First, add the imports, model constant, and basic client structure:
2103
2104 ```rust theme={null}
2105 use anyhow::{Context, Result, bail};
2106 use genai::Client;
2107 use genai::chat::{
2108 ChatMessage, ChatRequest, ChatResponse, ContentPart, Tool as GenaiTool, ToolResponse,
2109 };
2110 use rmcp::model::{CallToolRequestParam, Tool as McpTool};
2111 use rmcp::service::{RoleClient, RunningService, ServiceExt};
2112 use rmcp::transport::TokioChildProcess;
2113 use serde_json::Value;
2114 use tokio::io::{self, AsyncBufReadExt, BufReader};
2115 use tokio::process::Command;
2116
2117 const MODEL_ANTHROPIC: &str = "claude-sonnet-4-20250514";
2118
2119 struct MCPClient {
2120 anthropic: Client,
2121 session: Option<RunningService<RoleClient, ()>>,
2122 tools: Vec<GenaiTool>,
1846 end
1847end
1848
1849def cleanup
1850 @transport&.close
1851end
1852```
1853
1854### Main Entry Point
1855
1856Finally, we'll add the main execution logic:
1857
1858```ruby theme={null}
1859if ARGV.empty?
1860 puts "Usage: ruby client.rb <path_to_server_script>"
1861 exit 1
1862end
1863
1864client = MCPClient.new
1865
1866begin
1867 client.connect_to_server(ARGV[0])
1868
1869 api_key = ENV["ANTHROPIC_API_KEY"]
1870 if api_key.nil? || api_key.empty?
1871 puts <<~MESSAGE
1872 No ANTHROPIC_API_KEY found. To query these tools with Claude, set your API key:
1873 export ANTHROPIC_API_KEY=your-api-key-here
1874 MESSAGE
1875 exit
1876 end
1877
1878 client.chat_loop
1879rescue => e
1880 puts "Error: #{e.message}"
1881 exit 1
1882ensure
1883 client.cleanup
1884end
1885```
1886
1887You can find the complete `client.rb` file [here](https://github.com/modelcontextprotocol/quickstart-resources/blob/main/mcp-client-ruby/client.rb).
1888
1889## Key Components Explained
1890
1891### 1. Client Initialization
1892
1893* The `MCPClient` class initializes with nil references for lazy setup
1894* The Anthropic client is lazily initialized via the `anthropic_client` method
1895* Uses `dotenv` to load environment variables from `.env`
1896
1897### 2. Server Connection
1898
1899* Supports Ruby, Python, and Node.js servers
1900* Uses `File.extname` to determine the server script type
1901* Uses `MCP::Client::Stdio` for stdio transport
1902* Initializes the MCP client and lists available tools
1903
1904### 3. Query Processing
1905
1906* Maps MCP tools to Anthropic tool format (`name`, `description`, `input_schema`)
1907* Uses `Anthropic::Models::TextBlock` and `Anthropic::Models::ToolUseBlock` for pattern matching
1908* Builds assistant content once before iterating tool calls
1909* Executes tool calls via `@mcp_client.call_tool`
1910* Uses `chat` helper method to wrap Anthropic API calls
1911* Extracts tool result content with `result.dig("result", "content")`
1912* Passes tool results back to Claude for a final response
1913
1914### 4. Interactive Interface
1915
1916* Provides a simple command-line interface
1917* Handles user input and displays responses
1918* Skips empty queries
1919* Includes basic error handling
1920
1921### 5. Resource Management
1922
1923* Proper cleanup of the transport via `begin`...`ensure`
1924* Top-level `rescue` for error handling
1925* API key validation after server connection
1926
1927## Running the Client
1928
1929To run your client with any MCP server:
1930
1931```bash theme={null}
1932bundle exec ruby client.rb path/to/server.rb # ruby server
1933bundle exec ruby client.rb path/to/server.py # python server
1934bundle exec ruby client.rb path/to/build/index.js # node server
1935```
1936
1937> [!NOTE]
1938>
1939> If you're continuing [the weather tutorial from the server quickstart](https://github.com/modelcontextprotocol/quickstart-resources/tree/main/weather-server-ruby), your command might look something like this: `bundle exec ruby client.rb /path/to/weather-server-ruby/weather.rb`
1940
1941The client will:
1942
19431. Connect to the specified server
19442. List available tools
19453. Start an interactive chat session where you can:
1946 * Enter queries
1947 * See tool executions
1948 * Get responses from Claude
1949
1950## How It Works
1951
1952When you submit a query:
1953
19541. The client gets the list of available tools from the server
19552. Your query is sent to Claude along with tool descriptions
19563. Claude decides which tools (if any) to use
19574. The client executes any requested tool calls through the server
19585. Results are sent back to Claude
19596. Claude provides a natural language response
19607. The response is displayed to you
1961
1962## Best practices
1963
19641. **Error Handling**
1965 * Wrap tool calls in `begin`...`rescue` blocks
1966 * Provide meaningful error messages
1967 * Gracefully handle connection issues
1968
19692. **Resource Management**
1970 * Always close the transport when done
1971 * Use `begin`...`ensure` for proper cleanup
1972 * Handle server disconnections
1973
19743. **Security**
1975 * Store API keys securely in `.env`
1976 * Validate server responses
1977 * Be cautious with tool permissions
1978
1979## Troubleshooting
1980
1981### Server Path Issues
1982
1983* Double-check the path to your server script is correct
1984* Use the absolute path if the relative path isn't working
1985* For Windows users, make sure to use forward slashes (/) or escaped backslashes (\\) in the path
1986* Verify the server file has the correct extension (.py for Python, .js for Node.js, or .rb for Ruby)
1987
1988Example of correct path usage:
1989
1990```bash theme={null}
1991# Relative path
1992bundle exec ruby client.rb ./server/weather.rb
1993
1994# Absolute path
1995bundle exec ruby client.rb /Users/username/projects/mcp-server/weather.rb
1996
1997# Windows path (either format works)
1998bundle exec ruby client.rb C:/projects/mcp-server/weather.rb
1999bundle exec ruby client.rb C:\\projects\\mcp-server\\weather.rb
2000```
2001
2002### Response Timing
2003
2004* The first response might take up to 30 seconds to return
2005* This is normal and happens while:
2006 * The server initializes
2007 * Claude processes the query
2008 * Tools are being executed
2009* Subsequent responses are typically faster
2010* Don't interrupt the process during this initial waiting period
2011
2012### Common Error Messages
2013
2014If you see:
2015
2016* `Errno::ENOENT`: Check your server path and ensure the command (`ruby`, `python3`, `node`) is available
2017* `Connection refused`: Ensure the server is running and the path is correct
2018* `Tool execution failed`: Verify the tool's required environment variables are set
2019* `Anthropic::Errors::AuthenticationError`: Check your `.env` file has a valid `ANTHROPIC_API_KEY`
2020
2021**Rust**
2022
2023[You can find the complete code for this tutorial here.](https://github.com/modelcontextprotocol/quickstart-resources/tree/main/mcp-client-rust)
2024
2025## System Requirements
2026
2027Before starting, ensure your Linux system meets these requirements:
2028
2029* Latest stable version of [Rust and Cargo](https://www.rust-lang.org/tools/install)
2030* Anthropic API key (Claude)
2031* A Python, Node.js, or executable MCP server to connect to
2032
2033## Setting Up Your Environment
2034
2035First, create a new Rust project:
2036
2037```bash theme={null}
2038cargo new mcp-client-rust
2039cd mcp-client-rust
2040```
2041
2042Replace the contents of `Cargo.toml` with the following:
2043
2044```toml Cargo.toml theme={null}
2045[package]
2046name = "mcp-client-rust"
2047version = "0.1.0"
2048edition = "2024"
2049
2050[dependencies]
2051anyhow = "1.0.100"
2052genai = "0.4.2"
2053rmcp = { version = "0.8.0", features = ["server", "client", "transport-io", "transport-child-process"] }
2054tokio = { version = "1.47.1", features = ["full"] }
2055tracing = "0.1.41"
2056tracing-subscriber = { version = "0.3", features = ["env-filter"] }
2057serde_json = "1.0.128"
2058dotenvy = "0.15.7"
2059reqwest = "0.12.23"
2060```
2061
2062The [`rmcp`](https://github.com/modelcontextprotocol/rust-sdk) crate provides the Rust MCP SDK and child-process transport. This example uses the [`genai`](https://github.com/jeremychone/rust-genai) crate to send requests to Claude and represent tools in the model request.
2063
2064## Setting Up Your API Key
2065
2066You'll need an Anthropic API key from the [Anthropic Console](https://console.anthropic.com/settings/keys).
2067
2068Create a `.env` file to store it:
2069
2070```bash theme={null}
2071echo "ANTHROPIC_API_KEY=your-api-key-goes-here" > .env
2072```
2073
2074Add `.env` to your `.gitignore`:
2075
2076```bash theme={null}
2077echo ".env" >> .gitignore
2078```
2079
2080> [!WARNING]
2081>
2082> Make sure you keep your `ANTHROPIC_API_KEY` secure!
2083
2084## Creating the Client
2085
2086Open `src/main.rs` and replace its contents as you work through the following sections.
2087
2088### Imports and Client Structure
2089
2090First, add the imports, model constant, and basic client structure:
2091
2092```rust theme={null}
2093use anyhow::{Context, Result, bail};
2094use genai::Client;
2095use genai::chat::{
2096 ChatMessage, ChatRequest, ChatResponse, ContentPart, Tool as GenaiTool, ToolResponse,
2097};
2098use rmcp::model::{CallToolRequestParam, Tool as McpTool};
2099use rmcp::service::{RoleClient, RunningService, ServiceExt};
2100use rmcp::transport::TokioChildProcess;
2101use serde_json::Value;
2102use tokio::io::{self, AsyncBufReadExt, BufReader};
2103use tokio::process::Command;
2104
2105const MODEL_ANTHROPIC: &str = "claude-sonnet-4-20250514";
2106
2107struct MCPClient {
2108 anthropic: Client,
2109 session: Option<RunningService<RoleClient, ()>>,
2110 tools: Vec<GenaiTool>,
2111}
2112```
2113
2114The client keeps the model API client, the active MCP session, and the tools advertised by the connected server.
2115
2116### Client Initialization
2117
2118Next, initialize the model client and start without an MCP session or tools:
2119
2120```rust theme={null}
2121impl MCPClient {
2122 fn new() -> Result<Self> {
2123 Ok(MCPClient {
2124 anthropic: Client::default(),
2125 session: None,
2126 tools: Vec::new(),
2127 })
21232128 }
2124 ```
2125
2126 The client keeps the model API client, the active MCP session, and the tools advertised by the connected server.
2127
2128 ### Client Initialization
2129
2130 Next, initialize the model client and start without an MCP session or tools:
2131
2132 ```rust theme={null}
2133 impl MCPClient {
2134 fn new() -> Result<Self> {
2135 Ok(MCPClient {
2136 anthropic: Client::default(),
2137 session: None,
2138 tools: Vec::new(),
2139 })
2129
2130 // Additional methods will go here.
2131}
2132```
2133
2134`genai::Client::default()` reads the `ANTHROPIC_API_KEY` environment variable when it sends a request.
2135
2136### Server Connection Management
2137
2138Add this method inside the `impl MCPClient` block:
2139
2140```rust theme={null}
2141async fn connect_to_server(&mut self, server_args: &[String]) -> Result<()> {
2142 if self.session.is_some() {
2143 bail!("Client is already connected to a server");
2144 }
2145
2146 let mut command = Command::new(&server_args[0]);
2147 command.args(&server_args[1..]);
2148
2149 let process = TokioChildProcess::new(command)
2150 .with_context(|| format!("Failed to spawn server process for {:?}", server_args))?;
2151
2152 let session = ().serve(process).await?;
2153
2154 let rmcp_tools = session
2155 .list_all_tools()
2156 .await
2157 .context("Unable to list tools from server")?;
2158
2159 let tool_names: Vec<String> = rmcp_tools
2160 .iter()
2161 .map(|tool| tool.name.to_string())
2162 .collect();
2163
2164 println!("Connected to server with tools: {tool_names:?}");
2165
2166 self.tools = convert_tools(&rmcp_tools);
2167 self.session = Some(session);
2168 Ok(())
2169}
2170```
2171
2172This method:
2173
21741. Starts the server as a child process using the command and arguments supplied on the command line
21752. Establishes an MCP session over stdio
21763. Lists all tools advertised by the server
21774. Converts those tools into the format used in model requests
2178
2179### Converting MCP Tools
2180
2181Add this function outside the `impl MCPClient` block:
2182
2183```rust theme={null}
2184fn convert_tools(tools: &[McpTool]) -> Vec<GenaiTool> {
2185 tools
2186 .iter()
2187 .map(|tool| GenaiTool {
2188 name: tool.name.to_string(),
2189 description: tool.description.as_deref().map(str::to_string),
2190 schema: Some(Value::Object(tool.input_schema.as_ref().clone())),
2191 config: None,
2192 })
2193 .collect()
2194}
2195```
2196
2197MCP and model APIs describe tools with similar information but different Rust types. `convert_tools` maps each MCP tool's name, description, and input schema into a `genai` tool definition.
2198
2199### Sending Model Requests
2200
2201Add this helper method inside `impl MCPClient`:
2202
2203```rust theme={null}
2204async fn request_model(&self, chat_req: &ChatRequest) -> Result<ChatResponse> {
2205 let response = self
2206 .anthropic
2207 .exec_chat(MODEL_ANTHROPIC, chat_req.clone(), None)
2208 .await
2209 .context("Anthropic chat request failed")?;
2210
2211 Ok(response)
2212}
2213```
2214
2215This keeps model request handling in one place and adds useful context if the API request fails.
2216
2217### Query Processing Logic
2218
2219Now add the core query-processing method inside `impl MCPClient`:
2220
2221```rust theme={null}
2222async fn process_query(&mut self, query: &str) -> Result<String> {
2223 let session = self
2224 .session
2225 .as_ref()
2226 .context("Client is not connected to any server")?;
2227
2228 let mut messages = vec![ChatMessage::user(query)];
2229 let mut final_text = Vec::new();
2230
2231 // Initial Claude API call with tools
2232 let mut chat_req = ChatRequest::new(messages.clone()).with_tools(self.tools.clone());
2233 let mut chat_rsp = self.request_model(&chat_req).await?;
2234
2235 // Process response content - collect text and handle tool calls
2236 for text in chat_rsp.texts() {
2237 final_text.push(text.to_string());
2238 }
2239
2240 let tool_calls = chat_rsp.tool_calls();
2241 if !tool_calls.is_empty() {
2242 // Append assistant's response to message history
2243 messages.push(ChatMessage::assistant(chat_rsp.content.clone()));
2244
2245 // Execute each tool call and collect responses
2246 let mut tool_results = Vec::new();
2247 for tool_call in tool_calls {
2248 // Add information about the tool call to final text
2249 let tool_args_str = serde_json::to_string(&tool_call.fn_arguments)
2250 .unwrap_or_else(|_| "{}".to_string());
2251
2252 final_text.push(format!(
2253 "[Calling tool {} with args {}]",
2254 tool_call.fn_name, tool_args_str
2255 ));
2256
2257 // Query the MCP server
2258 let tool_result = session
2259 .call_tool(CallToolRequestParam {
2260 name: tool_call.fn_name.clone().into(),
2261 arguments: tool_call.fn_arguments.as_object().cloned(),
2262 })
2263 .await
2264 .with_context(|| format!("Tool call {} failed", tool_call.fn_name))?;
2265
2266 let payload = serde_json::to_string(&tool_result)
2267 .context("Failed to serialize tool result")?;
2268
2269 tool_results.push(ContentPart::ToolResponse(ToolResponse::new(
2270 tool_call.call_id.clone(),
2271 payload,
2272 )));
21402273 }
21412274
2142 // Additional methods will go here.
2143 }
2144 ```
2145
2146 `genai::Client::default()` reads the `ANTHROPIC_API_KEY` environment variable when it sends a request.
2147
2148 ### Server Connection Management
2149
2150 Add this method inside the `impl MCPClient` block:
2151
2152 ```rust theme={null}
2153 async fn connect_to_server(&mut self, server_args: &[String]) -> Result<()> {
2154 if self.session.is_some() {
2155 bail!("Client is already connected to a server");
2156 }
2157
2158 let mut command = Command::new(&server_args[0]);
2159 command.args(&server_args[1..]);
2160
2161 let process = TokioChildProcess::new(command)
2162 .with_context(|| format!("Failed to spawn server process for {:?}", server_args))?;
2163
2164 let session = ().serve(process).await?;
2165
2166 let rmcp_tools = session
2167 .list_all_tools()
2168 .await
2169 .context("Unable to list tools from server")?;
2170
2171 let tool_names: Vec<String> = rmcp_tools
2172 .iter()
2173 .map(|tool| tool.name.to_string())
2174 .collect();
2175
2176 println!("Connected to server with tools: {tool_names:?}");
2177
2178 self.tools = convert_tools(&rmcp_tools);
2179 self.session = Some(session);
2180 Ok(())
2181 }
2182 ```
2183
2184 This method:
2185
2186 1. Starts the server as a child process using the command and arguments supplied on the command line
2187 2. Establishes an MCP session over stdio
2188 3. Lists all tools advertised by the server
2189 4. Converts those tools into the format used in model requests
2190
2191 ### Converting MCP Tools
2192
2193 Add this function outside the `impl MCPClient` block:
2194
2195 ```rust theme={null}
2196 fn convert_tools(tools: &[McpTool]) -> Vec<GenaiTool> {
2197 tools
2198 .iter()
2199 .map(|tool| GenaiTool {
2200 name: tool.name.to_string(),
2201 description: tool.description.as_deref().map(str::to_string),
2202 schema: Some(Value::Object(tool.input_schema.as_ref().clone())),
2203 config: None,
2204 })
2205 .collect()
2206 }
2207 ```
2208
2209 MCP and model APIs describe tools with similar information but different Rust types. `convert_tools` maps each MCP tool's name, description, and input schema into a `genai` tool definition.
2210
2211 ### Sending Model Requests
2212
2213 Add this helper method inside `impl MCPClient`:
2214
2215 ```rust theme={null}
2216 async fn request_model(&self, chat_req: &ChatRequest) -> Result<ChatResponse> {
2217 let response = self
2218 .anthropic
2219 .exec_chat(MODEL_ANTHROPIC, chat_req.clone(), None)
2220 .await
2221 .context("Anthropic chat request failed")?;
2222
2223 Ok(response)
2224 }
2225 ```
2226
2227 This keeps model request handling in one place and adds useful context if the API request fails.
2228
2229 ### Query Processing Logic
2230
2231 Now add the core query-processing method inside `impl MCPClient`:
2232
2233 ```rust theme={null}
2234 async fn process_query(&mut self, query: &str) -> Result<String> {
2235 let session = self
2236 .session
2237 .as_ref()
2238 .context("Client is not connected to any server")?;
2239
2240 let mut messages = vec![ChatMessage::user(query)];
2241 let mut final_text = Vec::new();
2242
2243 // Initial Claude API call with tools
2244 let mut chat_req = ChatRequest::new(messages.clone()).with_tools(self.tools.clone());
2245 let mut chat_rsp = self.request_model(&chat_req).await?;
2246
2247 // Process response content - collect text and handle tool calls
2275 // Append tool responses to message history
2276 messages.push(ChatMessage::user(tool_results));
2277
2278 // Build the next request and query model
2279 chat_req = ChatRequest::new(messages.clone());
2280 chat_rsp = self.request_model(&chat_req).await?;
2281
2282 // Collect text from response
22482283 for text in chat_rsp.texts() {
22492284 final_text.push(text.to_string());
22502285 }
2251
2252 let tool_calls = chat_rsp.tool_calls();
2253 if !tool_calls.is_empty() {
2254 // Append assistant's response to message history
2255 messages.push(ChatMessage::assistant(chat_rsp.content.clone()));
2256
2257 // Execute each tool call and collect responses
2258 let mut tool_results = Vec::new();
2259 for tool_call in tool_calls {
2260 // Add information about the tool call to final text
2261 let tool_args_str = serde_json::to_string(&tool_call.fn_arguments)
2262 .unwrap_or_else(|_| "{}".to_string());
2263
2264 final_text.push(format!(
2265 "[Calling tool {} with args {}]",
2266 tool_call.fn_name, tool_args_str
2267 ));
2268
2269 // Query the MCP server
2270 let tool_result = session
2271 .call_tool(CallToolRequestParam {
2272 name: tool_call.fn_name.clone().into(),
2273 arguments: tool_call.fn_arguments.as_object().cloned(),
2274 })
2275 .await
2276 .with_context(|| format!("Tool call {} failed", tool_call.fn_name))?;
2277
2278 let payload = serde_json::to_string(&tool_result)
2279 .context("Failed to serialize tool result")?;
2280
2281 tool_results.push(ContentPart::ToolResponse(ToolResponse::new(
2282 tool_call.call_id.clone(),
2283 payload,
2284 )));
2285 }
2286
2287 // Append tool responses to message history
2288 messages.push(ChatMessage::user(tool_results));
2289
2290 // Build the next request and query model
2291 chat_req = ChatRequest::new(messages.clone());
2292 chat_rsp = self.request_model(&chat_req).await?;
2293
2294 // Collect text from response
2295 for text in chat_rsp.texts() {
2296 final_text.push(text.to_string());
2297 }
2286 }
2287
2288 Ok(final_text.join("\n"))
2289}
2290```
2291
2292The method first sends the user's query and available tools to Claude. When Claude requests tools, the client executes each request through the MCP session, sends the results back to Claude, and collects the final text response.
2293
2294### Interactive Chat Interface
2295
2296Add the interactive terminal loop inside `impl MCPClient`:
2297
2298```rust theme={null}
2299async fn chat_loop(&mut self) -> Result<()> {
2300 println!("\nMCP Client Started!");
2301 println!("Type your queries or 'quit' to exit.");
2302
2303 let mut stdin = BufReader::new(io::stdin());
2304 let mut input = String::new();
2305
2306 loop {
2307 print!("\nQuery: ");
2308 std::io::Write::flush(&mut std::io::stdout())?;
2309
2310 input.clear();
2311 if stdin.read_line(&mut input).await? == 0 {
2312 break; // EOF
22982313 }
22992314
2300 Ok(final_text.join("\n"))
2315 let query = input.trim();
2316 if query.eq_ignore_ascii_case("quit") {
2317 break;
2318 }
2319 if query.is_empty() {
2320 continue;
2321 }
2322
2323 match self.process_query(query).await {
2324 Ok(response) => println!("\n{}", response),
2325 Err(err) => println!("\nError: {}", err),
2326 }
23012327 }
2302 ```
2303
2304 The method first sends the user's query and available tools to Claude. When Claude requests tools, the client executes each request through the MCP session, sends the results back to Claude, and collects the final text response.
2305
2306 ### Interactive Chat Interface
2307
2308 Add the interactive terminal loop inside `impl MCPClient`:
2309
2310 ```rust theme={null}
2311 async fn chat_loop(&mut self) -> Result<()> {
2312 println!("\nMCP Client Started!");
2313 println!("Type your queries or 'quit' to exit.");
2314
2315 let mut stdin = BufReader::new(io::stdin());
2316 let mut input = String::new();
2317
2318 loop {
2319 print!("\nQuery: ");
2320 std::io::Write::flush(&mut std::io::stdout())?;
2321
2322 input.clear();
2323 if stdin.read_line(&mut input).await? == 0 {
2324 break; // EOF
2325 }
2326
2327 let query = input.trim();
2328 if query.eq_ignore_ascii_case("quit") {
2329 break;
2330 }
2331 if query.is_empty() {
2332 continue;
2333 }
2334
2335 match self.process_query(query).await {
2336 Ok(response) => println!("\n{}", response),
2337 Err(err) => println!("\nError: {}", err),
2338 }
2339 }
2340
2341 Ok(())
2328
2329 Ok(())
2330}
2331```
2332
2333The loop accepts queries until the user types `quit` or closes standard input. Query errors are printed without terminating the client.
2334
2335### Cleanup
2336
2337Add this method inside `impl MCPClient` to stop the MCP session and child process:
2338
2339```rust theme={null}
2340async fn cleanup(&mut self) -> Result<()> {
2341 if let Some(session) = self.session.take() {
2342 let _ = session.cancel().await;
23422343 }
2343 ```
2344
2345 The loop accepts queries until the user types `quit` or closes standard input. Query errors are printed without terminating the client.
2346
2347 ### Cleanup
2348
2349 Add this method inside `impl MCPClient` to stop the MCP session and child process:
2350
2351 ```rust theme={null}
2352 async fn cleanup(&mut self) -> Result<()> {
2353 if let Some(session) = self.session.take() {
2354 let _ = session.cancel().await;
2355 }
2356 Ok(())
2344 Ok(())
2345}
2346```
2347
2348### Main Entry Point
2349
2350Finally, add the asynchronous entry point outside the `impl MCPClient` block:
2351
2352```rust theme={null}
2353#[tokio::main]
2354async fn main() -> Result<()> {
2355 dotenvy::dotenv().context("Failed to load env file")?;
2356
2357 let mut args = std::env::args();
2358 let _ = args.next();
2359 let server_args: Vec<String> = args.collect();
2360
2361 if server_args.is_empty() {
2362 eprintln!("Usage: cargo run -- <server_script_or_binary> [args...]");
2363 std::process::exit(1);
23572364 }
2358 ```
2359
2360 ### Main Entry Point
2361
2362 Finally, add the asynchronous entry point outside the `impl MCPClient` block:
2363
2364 ```rust theme={null}
2365 #[tokio::main]
2366 async fn main() -> Result<()> {
2367 dotenvy::dotenv().context("Failed to load env file")?;
2368
2369 let mut args = std::env::args();
2370 let _ = args.next();
2371 let server_args: Vec<String> = args.collect();
2372
2373 if server_args.is_empty() {
2374 eprintln!("Usage: cargo run -- <server_script_or_binary> [args...]");
2375 std::process::exit(1);
2376 }
2377
2378 let mut client = MCPClient::new()?;
2379
2380 let result = async {
2381 client.connect_to_server(&server_args).await?;
2382 client.chat_loop().await
2383 }
2384 .await;
2385
2386 let cleanup_result = client.cleanup().await;
2387
2388 result?;
2389 cleanup_result?;
2390
2391 Ok(())
2365
2366 let mut client = MCPClient::new()?;
2367
2368 let result = async {
2369 client.connect_to_server(&server_args).await?;
2370 client.chat_loop().await
23922371 }
2393 ```
2394
2395 The entry point loads `.env`, treats all remaining command-line arguments as the server command, connects the client, starts the chat loop, and ensures cleanup runs before exiting.
2396
2397 ### Verify the Complete File
2398
2399 Before running the client, confirm the items in `src/main.rs` are placed at the correct scope:
2400
2401 * `new`, `connect_to_server`, `process_query`, `request_model`, `chat_loop`, and `cleanup` are methods inside the single `impl MCPClient` block.
2402 * `main` and `convert_tools` are functions outside the `impl MCPClient` block.
2403
2404 Rust does not require these items to appear in a particular order, but methods and free functions must be placed in the correct scope. Compare your file with the [complete `src/main.rs` example](https://github.com/modelcontextprotocol/quickstart-resources/blob/main/mcp-client-rust/src/main.rs), then check that it compiles:
2405
2406 ```bash theme={null}
2407 cargo fmt --check
2408 cargo check
2409 ```
2410
2411 ## Running the Client
2412
2413 Use `cargo run --` followed by the command you would normally use to start the MCP server:
2414
2415 ```bash theme={null}
2416 # Python server
2417 cargo run -- python path/to/server.py
2418
2419 # Node.js server
2420 cargo run -- node path/to/build/index.js
2421
2422 # Executable server
2423 cargo run -- path/to/server-binary
2424 ```
2425
2426 Running bare `cargo run` without a server command prints the usage message and exits.
2427
2428 <Note>
2429 If you're continuing [the weather tutorial from the server quickstart](https://github.com/modelcontextprotocol/quickstart-resources/tree/main/weather-server-rust), build the server first and then run a command similar to: `cargo run -- ../weather-server-rust/target/debug/weather`
2430 </Note>
2431
2432 The client will:
2433
2434 1. Start and connect to the specified MCP server
2435 2. List the tools available from that server
2436 3. Start an interactive chat session where you can:
2437 * Enter queries
2438 * See tool executions
2439 * Get responses from Claude
2440
2441 ## How It Works
2442
2443 When you submit a query:
2444
2445 1. The client sends your query and the server's available tools to Claude
2446 2. Claude decides which tools, if any, to use
2447 3. The client executes requested tools through the MCP session
2448 4. Tool results are sent back to Claude
2449 5. Claude provides a natural language response
2450 6. The response is displayed in the terminal
2451
2452 ## Best Practices
2453
2454 1. **Error Handling**
2455 * Add context to errors at process, MCP, model API, and serialization boundaries
2456 * Report individual query errors without terminating the interactive session
2457 * Validate server commands before running them
2458
2459 2. **Resource Management**
2460 * Always cancel the MCP session during cleanup
2461 * Ensure cleanup runs even when connection or chat-loop operations fail
2462 * Avoid starting a second server while a session is active
2463
2464 3. **Security**
2465 * Store API keys securely in `.env`
2466 * Review the tools exposed by a server before allowing model-driven calls
2467 * Connect only to servers and executable commands you trust
2468
2469 ## Troubleshooting
2470
2471 ### Server Command Issues
2472
2473 The arguments after `cargo run --` must form a complete command. Interpreted server scripts need their runtime:
2474
2475 ```bash theme={null}
2476 # Correct
2477 cargo run -- python ./server/weather.py
2478 cargo run -- node ./server/build/index.js
2479
2480 # Incorrect: a Python script is not necessarily executable by itself
2481 cargo run -- ./server/weather.py
2482 ```
2483
2484 If a command cannot be found, use its absolute path or verify it is available in your `PATH`.
2485
2486 ### Environment File Issues
2487
2488 If you see `Failed to load env file`, ensure `.env` exists in the directory where you run the client.
2489
2490 If the model request reports a missing API key, confirm that `.env` contains:
2491
2492 ```text theme={null}
2493 ANTHROPIC_API_KEY=your-api-key-goes-here
2494 ```
2495
2496 ### Tool and Response Errors
2497
2498 * `Unable to list tools from server`: Verify the server starts successfully and communicates over stdio
2499 * `Tool call ... failed`: Verify the server tool's required arguments and environment variables
2500 * `Failed to serialize tool result`: Inspect the server's response for unsupported or malformed content
2501 </Tab>
2502</Tabs>
2372 .await;
2373
2374 let cleanup_result = client.cleanup().await;
2375
2376 result?;
2377 cleanup_result?;
2378
2379 Ok(())
2380}
2381```
2382
2383The entry point loads `.env`, treats all remaining command-line arguments as the server command, connects the client, starts the chat loop, and ensures cleanup runs before exiting.
2384
2385### Verify the Complete File
2386
2387Before running the client, confirm the items in `src/main.rs` are placed at the correct scope:
2388
2389* `new`, `connect_to_server`, `process_query`, `request_model`, `chat_loop`, and `cleanup` are methods inside the single `impl MCPClient` block.
2390* `main` and `convert_tools` are functions outside the `impl MCPClient` block.
2391
2392Rust does not require these items to appear in a particular order, but methods and free functions must be placed in the correct scope. Compare your file with the [complete `src/main.rs` example](https://github.com/modelcontextprotocol/quickstart-resources/blob/main/mcp-client-rust/src/main.rs), then check that it compiles:
2393
2394```bash theme={null}
2395cargo fmt --check
2396cargo check
2397```
2398
2399## Running the Client
2400
2401Use `cargo run --` followed by the command you would normally use to start the MCP server:
2402
2403```bash theme={null}
2404# Python server
2405cargo run -- python path/to/server.py
2406
2407# Node.js server
2408cargo run -- node path/to/build/index.js
2409
2410# Executable server
2411cargo run -- path/to/server-binary
2412```
2413
2414Running bare `cargo run` without a server command prints the usage message and exits.
2415
2416> [!NOTE]
2417>
2418> If you're continuing [the weather tutorial from the server quickstart](https://github.com/modelcontextprotocol/quickstart-resources/tree/main/weather-server-rust), build the server first and then run a command similar to: `cargo run -- ../weather-server-rust/target/debug/weather`
2419
2420The client will:
2421
24221. Start and connect to the specified MCP server
24232. List the tools available from that server
24243. Start an interactive chat session where you can:
2425 * Enter queries
2426 * See tool executions
2427 * Get responses from Claude
2428
2429## How It Works
2430
2431When you submit a query:
2432
24331. The client sends your query and the server's available tools to Claude
24342. Claude decides which tools, if any, to use
24353. The client executes requested tools through the MCP session
24364. Tool results are sent back to Claude
24375. Claude provides a natural language response
24386. The response is displayed in the terminal
2439
2440## Best Practices
2441
24421. **Error Handling**
2443 * Add context to errors at process, MCP, model API, and serialization boundaries
2444 * Report individual query errors without terminating the interactive session
2445 * Validate server commands before running them
2446
24472. **Resource Management**
2448 * Always cancel the MCP session during cleanup
2449 * Ensure cleanup runs even when connection or chat-loop operations fail
2450 * Avoid starting a second server while a session is active
2451
24523. **Security**
2453 * Store API keys securely in `.env`
2454 * Review the tools exposed by a server before allowing model-driven calls
2455 * Connect only to servers and executable commands you trust
2456
2457## Troubleshooting
2458
2459### Server Command Issues
2460
2461The arguments after `cargo run --` must form a complete command. Interpreted server scripts need their runtime:
2462
2463```bash theme={null}
2464# Correct
2465cargo run -- python ./server/weather.py
2466cargo run -- node ./server/build/index.js
2467
2468# Incorrect: a Python script is not necessarily executable by itself
2469cargo run -- ./server/weather.py
2470```
2471
2472If a command cannot be found, use its absolute path or verify it is available in your `PATH`.
2473
2474### Environment File Issues
2475
2476If you see `Failed to load env file`, ensure `.env` exists in the directory where you run the client.
2477
2478If the model request reports a missing API key, confirm that `.env` contains:
2479
2480```text theme={null}
2481ANTHROPIC_API_KEY=your-api-key-goes-here
2482```
2483
2484### Tool and Response Errors
2485
2486* `Unable to list tools from server`: Verify the server starts successfully and communicates over stdio
2487* `Tool call ... failed`: Verify the server tool's required arguments and environment variables
2488* `Failed to serialize tool result`: Inspect the server's response for unsupported or malformed content
25032489
25042490## Next steps
25052491
2506<CardGroup cols={2}>
2507 <Card title="Example servers" icon="grid" href="/examples">
2508 Check out our gallery of official MCP servers and implementations
2509 </Card>
2510</CardGroup>
2492> **[Example servers](/examples)**
2493>
2494> Check out our gallery of official MCP servers and implementations
25112495
docs/2024-11-05/develop/build-server Changed · +2869 / -2937 lines
### Prerequisite knowledge ### Logging in MCP Servers ### Best Practices ### Quick Examples ### System requirements ### Set up your environment ## Building your server ### Importing packages and setting up the instance ### Helper functions ### Implementing tool execution ### Running the server ## Testing your server with Claude for Desktop ### Prerequisite knowledge ### Logging in MCP Servers ### Best Practices ### Quick Examples ### System requirements ### Set up your environment ## Building your server ### Importing packages and setting up the instance ### Helper functions ### Implementing tool execution ### Running the server ## Testing your server with Claude for Desktop ### Logging in MCP Servers ### Best Practices ### System requirements ### Set up your environment ## Building your server ### Weather Service ### Create your Boot Application ### Running the server ## Testing your server with Claude for Desktop ## Testing your server with Java client ### Create an MCP Client manually ### Use MCP Client Boot Starter ## More Java MCP Server examples ### Prerequisite knowledge ### Logging in MCP Servers ### Best Practices ### System requirements ### Set up your environment ## Building your server ### Setting up the instance ### Weather API helper functions ### Implementing tool execution ### Running the server ## Testing your server with Claude for Desktop ### Prerequisite knowledge ### Logging in MCP Servers ### Best Practices ### System requirements ### Set up your environment ## Building your server ### Weather API helper functions ### Running the server ## Testing your server with Claude for Desktop ### Prerequisite knowledge ### Logging in MCP Servers ### Best Practices ### Quick Examples ### System requirements ### Set up your environment ## Building your server ### Importing packages and setting up constants ### Helper methods ### Implementing tool execution ### Running the server ## Testing your server with Claude for Desktop ### Prerequisite knowledge ### Logging in MCP Servers ### Best Practices ### Quick Examples ### System requirements ### Set up your environment ## Building your server ### Importing packages and constants ### Data structures ### Helper functions ### Implementing the Weather server and tools ### Implementing the ServerHandler ### Running the server ## Testing your server with Claude for Desktop ### Prerequisite knowledge ### Logging in MCP Servers ### Best Practices ### Quick Examples ### System requirements ### Set up your environment ## Building your server ### Importing packages and constants ### Data structures ### Helper functions ### Implementing tool execution ### Running the server ## Testing your server with Claude for Desktop
The two sides of this change are more than 400 edits apart, too far apart to line up, so this is the differ's own diff of it and the words inside a line are not marked.
from line 8
88
99We'll build a server that exposes two tools: `get_alerts` and `get_forecast`. Then we'll connect the server to an MCP host (in this case, Claude for Desktop):
1010
11<Frame>
12 <img src="https://mintcdn.com/mcp/4ZXF1PrDkEaJvXpn/images/current-weather.png?fit=max&auto=format&n=4ZXF1PrDkEaJvXpn&q=85&s=dce7b2f8a06c20ba358e4bd2e75fa4c7" width="2780" height="1849" data-path="images/current-weather.png" />
13</Frame>
14
15<Note>
16 Servers can connect to any client. We've chosen Claude for Desktop here for simplicity, but we also have a guide on [building your own client](/docs/2024-11-05/develop/build-client).
17</Note>
11
12
13> [!NOTE]
14>
15> Servers can connect to any client. We've chosen Claude for Desktop here for simplicity, but we also have a guide on [building your own client](/docs/2024-11-05/develop/build-client).
1816
1917### Core MCP Concepts
2018
from line 24
2624
2725This tutorial will primarily focus on tools.
2826
29<Tabs>
30 <Tab title="Python">
31 Let's get started with building our weather server! [You can find the complete code for what we'll be building here.](https://github.com/modelcontextprotocol/quickstart-resources/tree/main/weather-server-python)
32
33 ### Prerequisite knowledge
34
35 This quickstart assumes you have familiarity with:
36
37 * Python
38 * LLMs like Claude
39
40 ### Logging in MCP Servers
41
42 When implementing MCP servers, be careful about how you handle logging:
43
44 **For STDIO-based servers:** Never write to stdout. Writing to stdout will corrupt the JSON-RPC messages and break your server. The `print()` function writes to stdout by default, but can be used safely with `file=sys.stderr`.
45
46 **For HTTP-based servers:** Standard output logging is fine since it doesn't interfere with HTTP responses.
47
48 ### Best Practices
49
50 * Use a logging library that writes to stderr or files.
51
52 ### Quick Examples
53
54 ```python theme={null}
55 import sys
56 import logging
57
58 # ❌ Bad (STDIO)
59 print("Processing request")
60
61 # ✅ Good (STDIO)
62 print("Processing request", file=sys.stderr)
63
64 # ✅ Good (STDIO)
65 logging.info("Processing request")
66 ```
67
68 ### System requirements
69
70 * Python 3.10 or higher installed.
71 * You must use the Python MCP SDK 1.2.0 or higher.
72
73 ### Set up your environment
74
75 First, let's install `uv` and set up our Python project and environment:
76
77 <CodeGroup>
78 ```bash macOS/Linux theme={null}
79 curl -LsSf https://astral.sh/uv/install.sh | sh
80 ```
81
82 ```powershell Windows theme={null}
83 powershell -ExecutionPolicy ByPass -c "irm https://astral.sh/uv/install.ps1 | iex"
84 ```
85 </CodeGroup>
86
87 Make sure to restart your terminal afterwards to ensure that the `uv` command gets picked up.
88
89 Now, let's create and set up our project:
90
91 <CodeGroup>
92 ```bash macOS/Linux theme={null}
93 # Create a new directory for our project
94 uv init weather
95 cd weather
96
97 # Create virtual environment and activate it
98 uv venv
99 source .venv/bin/activate
100
101 # Install dependencies
102 uv add "mcp[cli]" httpx
103
104 # Create our server file
105 touch weather.py
106 ```
107
108 ```powershell Windows theme={null}
109 # Create a new directory for our project
110 uv init weather
111 cd weather
112
113 # Create virtual environment and activate it
114 uv venv
115 .venv\Scripts\activate
116
117 # Install dependencies
118 uv add mcp[cli] httpx
119
120 # Create our server file
121 new-item weather.py
122 ```
123 </CodeGroup>
124
125 Now let's dive into building your server.
126
127 ## Building your server
128
129 ### Importing packages and setting up the instance
130
131 Add these to the top of your `weather.py`:
132
133 ```python theme={null}
134 from typing import Any
135
136 import httpx
137 from mcp.server.fastmcp import FastMCP
138
139 # Initialize FastMCP server
140 mcp = FastMCP("weather")
141
142 # Constants
143 NWS_API_BASE = "https://api.weather.gov"
144 USER_AGENT = "weather-app/1.0"
145 ```
146
147 The FastMCP class uses Python type hints and docstrings to automatically generate tool definitions, making it easy to create and maintain MCP tools.
148
149 ### Helper functions
150
151 Next, let's add our helper functions for querying and formatting the data from the National Weather Service API:
152
153 ```python theme={null}
154 async def make_nws_request(url: str) -> dict[str, Any] | None:
155 """Make a request to the NWS API with proper error handling."""
156 headers = {"User-Agent": USER_AGENT, "Accept": "application/geo+json"}
157 async with httpx.AsyncClient() as client:
158 try:
159 response = await client.get(url, headers=headers, timeout=30.0)
160 response.raise_for_status()
161 return response.json()
162 except Exception:
163 return None
164
165
166 def format_alert(feature: dict) -> str:
167 """Format an alert feature into a readable string."""
168 props = feature["properties"]
169 return f"""
170 Event: {props.get("event", "Unknown")}
171 Area: {props.get("areaDesc", "Unknown")}
172 Severity: {props.get("severity", "Unknown")}
173 Description: {props.get("description", "No description available")}
174 Instructions: {props.get("instruction", "No specific instructions provided")}
27**Python**
28
29Let's get started with building our weather server! [You can find the complete code for what we'll be building here.](https://github.com/modelcontextprotocol/quickstart-resources/tree/main/weather-server-python)
30
31### Prerequisite knowledge
32
33This quickstart assumes you have familiarity with:
34
35* Python
36* LLMs like Claude
37
38### Logging in MCP Servers
39
40When implementing MCP servers, be careful about how you handle logging:
41
42**For STDIO-based servers:** Never write to stdout. Writing to stdout will corrupt the JSON-RPC messages and break your server. The `print()` function writes to stdout by default, but can be used safely with `file=sys.stderr`.
43
44**For HTTP-based servers:** Standard output logging is fine since it doesn't interfere with HTTP responses.
45
46### Best Practices
47
48* Use a logging library that writes to stderr or files.
49
50### Quick Examples
51
52```python theme={null}
53import sys
54import logging
55
56# ❌ Bad (STDIO)
57print("Processing request")
58
59# ✅ Good (STDIO)
60print("Processing request", file=sys.stderr)
61
62# ✅ Good (STDIO)
63logging.info("Processing request")
64```
65
66### System requirements
67
68* Python 3.10 or higher installed.
69* You must use the Python MCP SDK 1.2.0 or higher.
70
71### Set up your environment
72
73First, let's install `uv` and set up our Python project and environment:
74
75```bash macOS/Linux theme={null}
76curl -LsSf https://astral.sh/uv/install.sh | sh
77```
78
79```powershell Windows theme={null}
80powershell -ExecutionPolicy ByPass -c "irm https://astral.sh/uv/install.ps1 | iex"
81```
82
83Make sure to restart your terminal afterwards to ensure that the `uv` command gets picked up.
84
85Now, let's create and set up our project:
86
87```bash macOS/Linux theme={null}
88# Create a new directory for our project
89uv init weather
90cd weather
91
92# Create virtual environment and activate it
93uv venv
94source .venv/bin/activate
95
96# Install dependencies
97uv add "mcp[cli]" httpx
98
99# Create our server file
100touch weather.py
101```
102
103```powershell Windows theme={null}
104# Create a new directory for our project
105uv init weather
106cd weather
107
108# Create virtual environment and activate it
109uv venv
110.venv\Scripts\activate
111
112# Install dependencies
113uv add mcp[cli] httpx
114
115# Create our server file
116new-item weather.py
117```
118
119Now let's dive into building your server.
120
121## Building your server
122
123### Importing packages and setting up the instance
124
125Add these to the top of your `weather.py`:
126
127```python theme={null}
128from typing import Any
129
130import httpx
131from mcp.server.fastmcp import FastMCP
132
133# Initialize FastMCP server
134mcp = FastMCP("weather")
135
136# Constants
137NWS_API_BASE = "https://api.weather.gov"
138USER_AGENT = "weather-app/1.0"
139```
140
141The FastMCP class uses Python type hints and docstrings to automatically generate tool definitions, making it easy to create and maintain MCP tools.
142
143### Helper functions
144
145Next, let's add our helper functions for querying and formatting the data from the National Weather Service API:
146
147```python theme={null}
148async def make_nws_request(url: str) -> dict[str, Any] | None:
149 """Make a request to the NWS API with proper error handling."""
150 headers = {"User-Agent": USER_AGENT, "Accept": "application/geo+json"}
151 async with httpx.AsyncClient() as client:
152 try:
153 response = await client.get(url, headers=headers, timeout=30.0)
154 response.raise_for_status()
155 return response.json()
156 except Exception:
157 return None
158
159
160def format_alert(feature: dict) -> str:
161 """Format an alert feature into a readable string."""
162 props = feature["properties"]
163 return f"""
164Event: {props.get("event", "Unknown")}
165Area: {props.get("areaDesc", "Unknown")}
166Severity: {props.get("severity", "Unknown")}
167Description: {props.get("description", "No description available")}
168Instructions: {props.get("instruction", "No specific instructions provided")}
169"""
170```
171
172### Implementing tool execution
173
174The tool execution handler is responsible for actually executing the logic of each tool. Let's add it:
175
176```python theme={null}
177@mcp.tool()
178async def get_alerts(state: str) -> str:
179 """Get weather alerts for a US state.
180
181 Args:
182 state: Two-letter US state code (e.g. CA, NY)
175183 """
176 ```
177
178 ### Implementing tool execution
179
180 The tool execution handler is responsible for actually executing the logic of each tool. Let's add it:
181
182 ```python theme={null}
183 @mcp.tool()
184 async def get_alerts(state: str) -> str:
185 """Get weather alerts for a US state.
186
187 Args:
188 state: Two-letter US state code (e.g. CA, NY)
184 url = f"{NWS_API_BASE}/alerts/active/area/{state}"
185 data = await make_nws_request(url)
186
187 if not data or "features" not in data:
188 return "Unable to fetch alerts or no alerts found."
189
190 if not data["features"]:
191 return "No active alerts for this state."
192
193 alerts = [format_alert(feature) for feature in data["features"]]
194 return "\n---\n".join(alerts)
195
196
197@mcp.tool()
198async def get_forecast(latitude: float, longitude: float) -> str:
199 """Get weather forecast for a location.
200
201 Args:
202 latitude: Latitude of the location
203 longitude: Longitude of the location
204 """
205 # First get the forecast grid endpoint
206 points_url = f"{NWS_API_BASE}/points/{latitude},{longitude}"
207 points_data = await make_nws_request(points_url)
208
209 if not points_data:
210 return "Unable to fetch forecast data for this location."
211
212 # Get the forecast URL from the points response
213 forecast_url = points_data["properties"]["forecast"]
214 forecast_data = await make_nws_request(forecast_url)
215
216 if not forecast_data:
217 return "Unable to fetch detailed forecast."
218
219 # Format the periods into a readable forecast
220 periods = forecast_data["properties"]["periods"]
221 forecasts = []
222 for period in periods[:5]: # Only show next 5 periods
223 forecast = f"""
224{period["name"]}:
225Temperature: {period["temperature"]}°{period["temperatureUnit"]}
226Wind: {period["windSpeed"]} {period["windDirection"]}
227Forecast: {period["detailedForecast"]}
228"""
229 forecasts.append(forecast)
230
231 return "\n---\n".join(forecasts)
232```
233
234### Running the server
235
236Finally, let's initialize and run the server:
237
238```python theme={null}
239def main():
240 # Initialize and run the server
241 mcp.run(transport="stdio")
242
243
244if __name__ == "__main__":
245 main()
246```
247
248Your server is complete! Run `uv run weather.py` to start the MCP server, which will listen for messages from MCP hosts.
249
250Let's now test your server from an existing MCP host, Claude for Desktop.
251
252## Testing your server with Claude for Desktop
253
254> [!NOTE]
255>
256> Claude for Desktop is not yet available on Linux. Linux users can proceed to the [Building a client](/docs/2024-11-05/develop/build-client) tutorial to build an MCP client that connects to the server we just built.
257
258First, make sure you have Claude for Desktop installed. [You can install the latest version
259here.](https://claude.ai/download) If you already have Claude for Desktop, **make sure it's updated to the latest version.**
260
261We'll need to configure Claude for Desktop for whichever MCP servers you want to use. To do this, open your Claude for Desktop App configuration at `~/Library/Application Support/Claude/claude_desktop_config.json` in a text editor. Make sure to create the file if it doesn't exist.
262
263For example, if you have [VS Code](https://code.visualstudio.com/) installed:
264
265```bash macOS/Linux theme={null}
266code ~/Library/Application\ Support/Claude/claude_desktop_config.json
267```
268
269```powershell Windows theme={null}
270code $env:AppData\Claude\claude_desktop_config.json
271```
272
273You'll then add your servers in the `mcpServers` key. The MCP UI elements will only show up in Claude for Desktop if at least one server is properly configured.
274
275In this case, we'll add our single weather server like so:
276
277```json macOS/Linux theme={null}
278{
279 "mcpServers": {
280 "weather": {
281 "command": "uv",
282 "args": [
283 "--directory",
284 "/ABSOLUTE/PATH/TO/PARENT/FOLDER/weather",
285 "run",
286 "weather.py"
287 ]
288 }
289 }
290}
291```
292
293```json Windows theme={null}
294{
295 "mcpServers": {
296 "weather": {
297 "command": "uv",
298 "args": [
299 "--directory",
300 "C:\\ABSOLUTE\\PATH\\TO\\PARENT\\FOLDER\\weather",
301 "run",
302 "weather.py"
303 ]
304 }
305 }
306}
307```
308
309> [!WARNING]
310>
311> You may need to put the full path to the `uv` executable in the `command` field. You can get this by running `which uv` on macOS/Linux or `where uv` on Windows.
312
313> [!NOTE]
314>
315> Make sure you pass in the absolute path to your server. You can get this by running `pwd` on macOS/Linux or `cd` on Windows Command Prompt. On Windows, remember to use double backslashes (`\\`) or forward slashes (`/`) in the JSON path.
316
317This tells Claude for Desktop:
318
3191. There's an MCP server named "weather"
3202. To launch it by running `uv --directory /ABSOLUTE/PATH/TO/PARENT/FOLDER/weather run weather.py`
321
322Save the file, and restart **Claude for Desktop**.
323
324**TypeScript**
325
326Let's get started with building our weather server! [You can find the complete code for what we'll be building here.](https://github.com/modelcontextprotocol/quickstart-resources/tree/main/weather-server-typescript)
327
328### Prerequisite knowledge
329
330This quickstart assumes you have familiarity with:
331
332* TypeScript
333* LLMs like Claude
334
335### Logging in MCP Servers
336
337When implementing MCP servers, be careful about how you handle logging:
338
339**For STDIO-based servers:** Never use `console.log()`, as it writes to standard output (stdout) by default. Writing to stdout will corrupt the JSON-RPC messages and break your server.
340
341**For HTTP-based servers:** Standard output logging is fine since it doesn't interfere with HTTP responses.
342
343### Best Practices
344
345* Use `console.error()` which writes to stderr, or use a logging library that writes to stderr or files.
346
347### Quick Examples
348
349```javascript theme={null}
350// ❌ Bad (STDIO)
351console.log("Server started");
352
353// ✅ Good (STDIO)
354console.error("Server started"); // stderr is safe
355```
356
357### System requirements
358
359For TypeScript, make sure you have the latest version of Node installed.
360
361### Set up your environment
362
363First, let's install Node.js and npm if you haven't already. You can download them from [nodejs.org](https://nodejs.org/).
364Verify your Node.js installation:
365
366```bash theme={null}
367node --version
368npm --version
369```
370
371For this tutorial, you'll need Node.js version 16 or higher.
372
373Now, let's create and set up our project:
374
375```bash macOS/Linux theme={null}
376# Create a new directory for our project
377mkdir weather
378cd weather
379
380# Initialize a new npm project
381npm init -y
382
383# Install dependencies
384npm install @modelcontextprotocol/sdk zod@3
385npm install -D @types/node typescript
386
387# Create our files
388mkdir src
389touch src/index.ts
390```
391
392```powershell Windows theme={null}
393# Create a new directory for our project
394md weather
395cd weather
396
397# Initialize a new npm project
398npm init -y
399
400# Install dependencies
401npm install @modelcontextprotocol/sdk zod@3
402npm install -D @types/node typescript
403
404# Create our files
405md src
406new-item src\index.ts
407```
408
409Update your package.json to add type: "module" and a build script:
410
411```json package.json theme={null}
412{
413 "type": "module",
414 "bin": {
415 "weather": "./build/index.js"
416 },
417 "scripts": {
418 "build": "tsc && chmod 755 build/index.js"
419 },
420 "files": ["build"]
421}
422```
423
424Create a `tsconfig.json` in the root of your project:
425
426```json tsconfig.json theme={null}
427{
428 "compilerOptions": {
429 "target": "ES2022",
430 "module": "Node16",
431 "moduleResolution": "Node16",
432 "outDir": "./build",
433 "rootDir": "./src",
434 "strict": true,
435 "esModuleInterop": true,
436 "skipLibCheck": true,
437 "forceConsistentCasingInFileNames": true
438 },
439 "include": ["src/**/*"],
440 "exclude": ["node_modules"]
441}
442```
443
444Now let's dive into building your server.
445
446## Building your server
447
448### Importing packages and setting up the instance
449
450Add these to the top of your `src/index.ts`:
451
452```typescript theme={null}
453import { McpServer } from "@modelcontextprotocol/sdk/server/mcp.js";
454import { StdioServerTransport } from "@modelcontextprotocol/sdk/server/stdio.js";
455import { z } from "zod";
456
457const NWS_API_BASE = "https://api.weather.gov";
458const USER_AGENT = "weather-app/1.0";
459
460// Create server instance
461const server = new McpServer({
462 name: "weather",
463 version: "1.0.0",
464});
465```
466
467### Helper functions
468
469Next, let's add our helper functions for querying and formatting the data from the National Weather Service API:
470
471```typescript theme={null}
472// Helper function for making NWS API requests
473async function makeNWSRequest<T>(url: string): Promise<T | null> {
474 const headers = {
475 "User-Agent": USER_AGENT,
476 Accept: "application/geo+json",
477 };
478
479 try {
480 const response = await fetch(url, { headers });
481 if (!response.ok) {
482 throw new Error(`HTTP error! status: ${response.status}`);
483 }
484 return (await response.json()) as T;
485 } catch (error) {
486 console.error("Error making NWS request:", error);
487 return null;
488 }
489}
490
491interface AlertFeature {
492 properties: {
493 event?: string;
494 areaDesc?: string;
495 severity?: string;
496 status?: string;
497 headline?: string;
498 };
499}
500
501// Format alert data
502function formatAlert(feature: AlertFeature): string {
503 const props = feature.properties;
504 return [
505 `Event: ${props.event || "Unknown"}`,
506 `Area: ${props.areaDesc || "Unknown"}`,
507 `Severity: ${props.severity || "Unknown"}`,
508 `Status: ${props.status || "Unknown"}`,
509 `Headline: ${props.headline || "No headline"}`,
510 "---",
511 ].join("\n");
512}
513
514interface ForecastPeriod {
515 name?: string;
516 temperature?: number;
517 temperatureUnit?: string;
518 windSpeed?: string;
519 windDirection?: string;
520 shortForecast?: string;
521}
522
523interface AlertsResponse {
524 features: AlertFeature[];
525}
526
527interface PointsResponse {
528 properties: {
529 forecast?: string;
530 };
531}
532
533interface ForecastResponse {
534 properties: {
535 periods: ForecastPeriod[];
536 };
537}
538```
539
540### Implementing tool execution
541
542The tool execution handler is responsible for actually executing the logic of each tool. Let's add it:
543
544```typescript theme={null}
545// Register weather tools
546
547server.registerTool(
548 "get_alerts",
549 {
550 description: "Get weather alerts for a state",
551 inputSchema: {
552 state: z
553 .string()
554 .length(2)
555 .describe("Two-letter state code (e.g. CA, NY)"),
556 },
557 },
558 async ({ state }) => {
559 const stateCode = state.toUpperCase();
560 const alertsUrl = `${NWS_API_BASE}/alerts?area=${stateCode}`;
561 const alertsData = await makeNWSRequest<AlertsResponse>(alertsUrl);
562
563 if (!alertsData) {
564 return {
565 content: [
566 {
567 type: "text",
568 text: "Failed to retrieve alerts data",
569 },
570 ],
571 };
572 }
573
574 const features = alertsData.features || [];
575 if (!features.length) {
576 return {
577 content: [
578 {
579 type: "text",
580 text: `No active alerts for ${stateCode}`,
581 },
582 ],
583 };
584 }
585
586 const formattedAlerts = features.map(formatAlert);
587 const alertsText = `Active alerts for ${stateCode}:\n\n${formattedAlerts.join("\n")}`;
588
589 return {
590 content: [
591 {
592 type: "text",
593 text: alertsText,
594 },
595 ],
596 };
597 },
598);
599
600server.registerTool(
601 "get_forecast",
602 {
603 description: "Get weather forecast for a location",
604 inputSchema: {
605 latitude: z
606 .number()
607 .min(-90)
608 .max(90)
609 .describe("Latitude of the location"),
610 longitude: z
611 .number()
612 .min(-180)
613 .max(180)
614 .describe("Longitude of the location"),
615 },
616 },
617 async ({ latitude, longitude }) => {
618 // Get grid point data
619 const pointsUrl = `${NWS_API_BASE}/points/${latitude.toFixed(4)},${longitude.toFixed(4)}`;
620 const pointsData = await makeNWSRequest<PointsResponse>(pointsUrl);
621
622 if (!pointsData) {
623 return {
624 content: [
625 {
626 type: "text",
627 text: `Failed to retrieve grid point data for coordinates: ${latitude}, ${longitude}. This location may not be supported by the NWS API (only US locations are supported).`,
628 },
629 ],
630 };
631 }
632
633 const forecastUrl = pointsData.properties?.forecast;
634 if (!forecastUrl) {
635 return {
636 content: [
637 {
638 type: "text",
639 text: "Failed to get forecast URL from grid point data",
640 },
641 ],
642 };
643 }
644
645 // Get forecast data
646 const forecastData = await makeNWSRequest<ForecastResponse>(forecastUrl);
647 if (!forecastData) {
648 return {
649 content: [
650 {
651 type: "text",
652 text: "Failed to retrieve forecast data",
653 },
654 ],
655 };
656 }
657
658 const periods = forecastData.properties?.periods || [];
659 if (periods.length === 0) {
660 return {
661 content: [
662 {
663 type: "text",
664 text: "No forecast periods available",
665 },
666 ],
667 };
668 }
669
670 // Format forecast periods
671 const formattedForecast = periods.map((period: ForecastPeriod) =>
672 [
673 `${period.name || "Unknown"}:`,
674 `Temperature: ${period.temperature || "Unknown"}°${period.temperatureUnit || "F"}`,
675 `Wind: ${period.windSpeed || "Unknown"} ${period.windDirection || ""}`,
676 `${period.shortForecast || "No forecast available"}`,
677 "---",
678 ].join("\n"),
679 );
680
681 const forecastText = `Forecast for ${latitude}, ${longitude}:\n\n${formattedForecast.join("\n")}`;
682
683 return {
684 content: [
685 {
686 type: "text",
687 text: forecastText,
688 },
689 ],
690 };
691 },
692);
693```
694
695### Running the server
696
697Finally, implement the main function to run the server:
698
699```typescript theme={null}
700async function main() {
701 const transport = new StdioServerTransport();
702 await server.connect(transport);
703 console.error("Weather MCP Server running on stdio");
704}
705
706main().catch((error) => {
707 console.error("Fatal error in main():", error);
708 process.exit(1);
709});
710```
711
712Make sure to run `npm run build` to build your server! This is a very important step in getting your server to connect.
713
714Let's now test your server from an existing MCP host, Claude for Desktop.
715
716## Testing your server with Claude for Desktop
717
718> [!NOTE]
719>
720> Claude for Desktop is not yet available on Linux. Linux users can proceed to the [Building a client](/docs/2024-11-05/develop/build-client) tutorial to build an MCP client that connects to the server we just built.
721
722First, make sure you have Claude for Desktop installed. [You can install the latest version
723here.](https://claude.ai/download) If you already have Claude for Desktop, **make sure it's updated to the latest version.**
724
725We'll need to configure Claude for Desktop for whichever MCP servers you want to use. To do this, open your Claude for Desktop App configuration at `~/Library/Application Support/Claude/claude_desktop_config.json` in a text editor. Make sure to create the file if it doesn't exist.
726
727For example, if you have [VS Code](https://code.visualstudio.com/) installed:
728
729```bash macOS/Linux theme={null}
730code ~/Library/Application\ Support/Claude/claude_desktop_config.json
731```
732
733```powershell Windows theme={null}
734code $env:AppData\Claude\claude_desktop_config.json
735```
736
737You'll then add your servers in the `mcpServers` key. The MCP UI elements will only show up in Claude for Desktop if at least one server is properly configured.
738
739In this case, we'll add our single weather server like so:
740
741```json macOS/Linux theme={null}
742{
743 "mcpServers": {
744 "weather": {
745 "command": "node",
746 "args": ["/ABSOLUTE/PATH/TO/PARENT/FOLDER/weather/build/index.js"]
747 }
748 }
749}
750```
751
752```json Windows theme={null}
753{
754 "mcpServers": {
755 "weather": {
756 "command": "node",
757 "args": ["C:\\PATH\\TO\\PARENT\\FOLDER\\weather\\build\\index.js"]
758 }
759 }
760}
761```
762
763This tells Claude for Desktop:
764
7651. There's an MCP server named "weather"
7662. Launch it by running `node /ABSOLUTE/PATH/TO/PARENT/FOLDER/weather/build/index.js`
767
768Save the file, and restart **Claude for Desktop**.
769
770**Java**
771
772> [!NOTE]
773>
774> This is a quickstart demo based on Spring AI MCP auto-configuration and boot starters.
775> To learn how to create sync and async MCP Servers, manually, consult the [Java SDK Server](https://java.sdk.modelcontextprotocol.io/) documentation.
776
777Let's get started with building our weather server!
778[You can find the complete code for what we'll be building here.](https://github.com/spring-projects/spring-ai-examples/tree/main/model-context-protocol/weather/starter-stdio-server)
779
780For more information, see the [MCP Server Boot Starter](https://docs.spring.io/spring-ai/reference/api/mcp/mcp-server-boot-starter-docs.html) reference documentation.
781For manual MCP Server implementation, refer to the [MCP Server Java SDK documentation](https://java.sdk.modelcontextprotocol.io/).
782
783### Logging in MCP Servers
784
785When implementing MCP servers, be careful about how you handle logging:
786
787**For STDIO-based servers:** Never use `System.out.println()` or `System.out.print()`, as they write to standard output (stdout). Writing to stdout will corrupt the JSON-RPC messages and break your server.
788
789**For HTTP-based servers:** Standard output logging is fine since it doesn't interfere with HTTP responses.
790
791### Best Practices
792
793* Use a logging library that writes to stderr or files.
794* Ensure any configured logging library will not write to stdout.
795
796### System requirements
797
798* Java 17 or higher installed.
799* [Spring Boot 3.3.x](https://docs.spring.io/spring-boot/installing.html) or higher
800
801### Set up your environment
802
803Use the [Spring Initializer](https://start.spring.io/) to bootstrap the project.
804
805You will need to add the following dependencies:
806
807```xml Maven theme={null}
808<dependencies>
809 <dependency>
810 <groupId>org.springframework.ai</groupId>
811 <artifactId>spring-ai-starter-mcp-server</artifactId>
812 </dependency>
813
814 <dependency>
815 <groupId>org.springframework</groupId>
816 <artifactId>spring-web</artifactId>
817 </dependency>
818</dependencies>
819```
820
821```groovy Gradle theme={null}
822dependencies {
823 implementation platform("org.springframework.ai:spring-ai-starter-mcp-server")
824 implementation platform("org.springframework:spring-web")
825}
826```
827
828Then configure your application by setting the application properties:
829
830```bash application.properties theme={null}
831spring.main.bannerMode=off
832logging.pattern.console=
833```
834
835```yaml application.yml theme={null}
836logging:
837 pattern:
838 console:
839spring:
840 main:
841 banner-mode: off
842```
843
844The [Server Configuration Properties](https://docs.spring.io/spring-ai/reference/api/mcp/mcp-server-boot-starter-docs.html#_configuration_properties) documents all available properties.
845
846Now let's dive into building your server.
847
848## Building your server
849
850### Weather Service
851
852Let's implement a [WeatherService.java](https://github.com/spring-projects/spring-ai-examples/blob/main/model-context-protocol/weather/starter-stdio-server/src/main/java/org/springframework/ai/mcp/sample/server/WeatherService.java) that uses a REST client to query the data from the National Weather Service API:
853
854```java theme={null}
855@Service
856public class WeatherService {
857
858 private final RestClient restClient;
859
860 public WeatherService() {
861 this.restClient = RestClient.builder()
862 .baseUrl("https://api.weather.gov")
863 .defaultHeader("Accept", "application/geo+json")
864 .defaultHeader("User-Agent", "WeatherApiClient/1.0 ([email protected])")
865 .build();
866 }
867
868 @Tool(description = "Get weather forecast for a specific latitude/longitude")
869 public String getWeatherForecastByLocation(
870 double latitude, // Latitude coordinate
871 double longitude // Longitude coordinate
872 ) {
873 // Returns detailed forecast including:
874 // - Temperature and unit
875 // - Wind speed and direction
876 // - Detailed forecast description
877 }
878
879 @Tool(description = "Get weather alerts for a US state")
880 public String getAlerts(
881 @ToolParam(description = "Two-letter US state code (e.g. CA, NY)") String state
882 ) {
883 // Returns active alerts including:
884 // - Event type
885 // - Affected area
886 // - Severity
887 // - Description
888 // - Safety instructions
889 }
890
891 // ......
892}
893```
894
895The `@Service` annotation will auto-register the service in your application context.
896The Spring AI `@Tool` annotation makes it easy to create and maintain MCP tools.
897
898The auto-configuration will automatically register these tools with the MCP server.
899
900### Create your Boot Application
901
902```java theme={null}
903@SpringBootApplication
904public class McpServerApplication {
905
906 public static void main(String[] args) {
907 SpringApplication.run(McpServerApplication.class, args);
908 }
909
910 @Bean
911 public ToolCallbackProvider weatherTools(WeatherService weatherService) {
912 return MethodToolCallbackProvider.builder().toolObjects(weatherService).build();
913 }
914}
915```
916
917Uses the `MethodToolCallbackProvider` utils to convert the `@Tools` into actionable callbacks used by the MCP server.
918
919### Running the server
920
921Finally, let's build the server:
922
923```bash theme={null}
924./mvnw clean install
925```
926
927This will generate an `mcp-weather-stdio-server-0.0.1-SNAPSHOT.jar` file within the `target` folder.
928
929Let's now test your server from an existing MCP host, Claude for Desktop.
930
931## Testing your server with Claude for Desktop
932
933> [!NOTE]
934>
935> Claude for Desktop is not yet available on Linux.
936
937First, make sure you have Claude for Desktop installed.
938[You can install the latest version here.](https://claude.ai/download) If you already have Claude for Desktop, **make sure it's updated to the latest version.**
939
940We'll need to configure Claude for Desktop for whichever MCP servers you want to use.
941To do this, open your Claude for Desktop App configuration at `~/Library/Application Support/Claude/claude_desktop_config.json` in a text editor.
942Make sure to create the file if it doesn't exist.
943
944For example, if you have [VS Code](https://code.visualstudio.com/) installed:
945
946```bash macOS/Linux theme={null}
947code ~/Library/Application\ Support/Claude/claude_desktop_config.json
948```
949
950```powershell Windows theme={null}
951code $env:AppData\Claude\claude_desktop_config.json
952```
953
954You'll then add your servers in the `mcpServers` key.
955The MCP UI elements will only show up in Claude for Desktop if at least one server is properly configured.
956
957In this case, we'll add our single weather server like so:
958
959```json macOS/Linux theme={null}
960{
961 "mcpServers": {
962 "spring-ai-mcp-weather": {
963 "command": "java",
964 "args": [
965 "-Dspring.ai.mcp.server.stdio=true",
966 "-jar",
967 "/ABSOLUTE/PATH/TO/PARENT/FOLDER/mcp-weather-stdio-server-0.0.1-SNAPSHOT.jar"
968 ]
969 }
970 }
971}
972```
973
974```json Windows theme={null}
975{
976 "mcpServers": {
977 "spring-ai-mcp-weather": {
978 "command": "java",
979 "args": [
980 "-Dspring.ai.mcp.server.transport=STDIO",
981 "-jar",
982 "C:\\ABSOLUTE\\PATH\\TO\\PARENT\\FOLDER\\weather\\mcp-weather-stdio-server-0.0.1-SNAPSHOT.jar"
983 ]
984 }
985 }
986}
987```
988
989> [!NOTE]
990>
991> Make sure you pass in the absolute path to your server.
992
993This tells Claude for Desktop:
994
9951. There's an MCP server named "my-weather-server"
9962. To launch it by running `java -jar /ABSOLUTE/PATH/TO/PARENT/FOLDER/mcp-weather-stdio-server-0.0.1-SNAPSHOT.jar`
997
998Save the file, and restart **Claude for Desktop**.
999
1000## Testing your server with Java client
1001
1002### Create an MCP Client manually
1003
1004Use the `McpClient` to connect to the server:
1005
1006```java theme={null}
1007var stdioParams = ServerParameters.builder("java")
1008 .args("-jar", "/ABSOLUTE/PATH/TO/PARENT/FOLDER/mcp-weather-stdio-server-0.0.1-SNAPSHOT.jar")
1009 .build();
1010
1011var stdioTransport = new StdioClientTransport(stdioParams);
1012
1013var mcpClient = McpClient.sync(stdioTransport).build();
1014
1015mcpClient.initialize();
1016
1017ListToolsResult toolsList = mcpClient.listTools();
1018
1019CallToolResult weather = mcpClient.callTool(
1020 new CallToolRequest("getWeatherForecastByLocation",
1021 Map.of("latitude", "47.6062", "longitude", "-122.3321")));
1022
1023CallToolResult alert = mcpClient.callTool(
1024 new CallToolRequest("getAlerts", Map.of("state", "NY")));
1025
1026mcpClient.closeGracefully();
1027```
1028
1029### Use MCP Client Boot Starter
1030
1031Create a new boot starter application using the `spring-ai-starter-mcp-client` dependency:
1032
1033```xml theme={null}
1034<dependency>
1035 <groupId>org.springframework.ai</groupId>
1036 <artifactId>spring-ai-starter-mcp-client</artifactId>
1037</dependency>
1038```
1039
1040and set the `spring.ai.mcp.client.stdio.servers-configuration` property to point to your `claude_desktop_config.json`.
1041You can reuse the existing Anthropic Desktop configuration:
1042
1043```properties theme={null}
1044spring.ai.mcp.client.stdio.servers-configuration=file:PATH/TO/claude_desktop_config.json
1045```
1046
1047When you start your client application, the auto-configuration will automatically create MCP clients from the claude\_desktop\_config.json.
1048
1049For more information, see the [MCP Client Boot Starters](https://docs.spring.io/spring-ai/reference/api/mcp/mcp-client-boot-starter-docs.html) reference documentation.
1050
1051## More Java MCP Server examples
1052
1053The [starter-webflux-server](https://github.com/spring-projects/spring-ai-examples/tree/main/model-context-protocol/weather/starter-webflux-server) demonstrates how to create an MCP server using SSE transport.
1054It showcases how to define and register MCP Tools, Resources, and Prompts, using the Spring Boot's auto-configuration capabilities.
1055
1056**Kotlin**
1057
1058Let's get started with building our weather server! [You can find the complete code for what we'll be building here.](https://github.com/modelcontextprotocol/kotlin-sdk/tree/main/samples/weather-stdio-server)
1059
1060### Prerequisite knowledge
1061
1062This quickstart assumes you have familiarity with:
1063
1064* Kotlin
1065* LLMs like Claude
1066
1067### Logging in MCP Servers
1068
1069When implementing MCP servers, be careful about how you handle logging:
1070
1071**For STDIO-based servers:** Never use `println()`, as it writes to standard output (stdout) by default. Writing to stdout will corrupt the JSON-RPC messages and break your server.
1072
1073**For HTTP-based servers:** Standard output logging is fine since it doesn't interfere with HTTP responses.
1074
1075### Best Practices
1076
1077* Use a logging library that writes to stderr or files.
1078
1079### System requirements
1080
1081* JDK 11 or higher installed.
1082
1083### Set up your environment
1084
1085First, let's install `java` and `gradle` if you haven't already.
1086You can download `java` from [official Oracle JDK website](https://www.oracle.com/java/technologies/downloads/).
1087Verify your `java` installation:
1088
1089```bash theme={null}
1090java --version
1091```
1092
1093Now, let's create and set up your project:
1094
1095```bash macOS/Linux theme={null}
1096# Create a new directory for our project
1097mkdir weather
1098cd weather
1099
1100# Initialize a new kotlin project
1101gradle init
1102```
1103
1104```powershell Windows theme={null}
1105# Create a new directory for our project
1106md weather
1107cd weather
1108
1109# Initialize a new kotlin project
1110gradle init
1111```
1112
1113After running `gradle init`, select **Application** as the project type, **Kotlin** as the programming language.
1114
1115Alternatively, you can create a Kotlin application using the [IntelliJ IDEA project wizard](https://kotlinlang.org/docs/jvm-get-started.html).
1116
1117After creating the project, replace the contents of your `build.gradle.kts` with:
1118
1119```kotlin build.gradle.kts theme={null}
1120// Check latest versions at https://github.com/modelcontextprotocol/kotlin-sdk/releases
1121val mcpVersion = "0.9.0"
1122val ktorVersion = "3.2.3"
1123val slf4jVersion = "2.0.17"
1124
1125plugins {
1126 kotlin("jvm") version "2.3.20"
1127 kotlin("plugin.serialization") version "2.3.20"
1128 id("com.gradleup.shadow") version "8.3.9"
1129 application
1130}
1131
1132application {
1133 mainClass.set("MainKt")
1134}
1135
1136dependencies {
1137 implementation("io.modelcontextprotocol:kotlin-sdk:$mcpVersion")
1138 implementation("io.ktor:ktor-client-content-negotiation:$ktorVersion")
1139 implementation("io.ktor:ktor-serialization-kotlinx-json:$ktorVersion")
1140 implementation("io.ktor:ktor-client-cio:$ktorVersion")
1141 implementation("org.slf4j:slf4j-simple:$slf4jVersion")
1142}
1143```
1144
1145Verify that everything is set up correctly:
1146
1147```bash theme={null}
1148./gradlew build
1149```
1150
1151Now let's dive into building your server.
1152
1153## Building your server
1154
1155### Setting up the instance
1156
1157Add a server initialization function:
1158
1159```kotlin theme={null}
1160fun runMcpServer() {
1161 val server = Server(
1162 Implementation(
1163 name = "weather",
1164 version = "1.0.0",
1165 ),
1166 ServerOptions(
1167 capabilities = ServerCapabilities(tools = ServerCapabilities.Tools(listChanged = true)),
1168 ),
1169 )
1170
1171 // register tools on server here
1172
1173 val transport = StdioServerTransport(
1174 System.`in`.asInput(),
1175 System.out.asSink().buffered(),
1176 )
1177
1178 runBlocking {
1179 val session = server.createSession(transport)
1180 val done = Job()
1181 session.onClose {
1182 done.complete()
1183 }
1184 done.join()
1185 }
1186}
1187```
1188
1189### Weather API helper functions
1190
1191Next, let's add functions and data classes for querying and converting responses from the National Weather Service API:
1192
1193```kotlin theme={null}
1194val httpClient = HttpClient(CIO) {
1195 defaultRequest {
1196 url("https://api.weather.gov")
1197 headers {
1198 append("Accept", "application/geo+json")
1199 append("User-Agent", "WeatherApiClient/1.0")
1200 }
1201 contentType(ContentType.Application.Json)
1202 }
1203 install(ContentNegotiation) {
1204 json(Json { ignoreUnknownKeys = true })
1205 }
1206}
1207
1208// Extension function to fetch weather alerts for a given state
1209suspend fun HttpClient.getAlerts(state: String): List<String> {
1210 val alerts = this.get("/alerts/active/area/$state").body<AlertsResponse>()
1211 return alerts.features.map { feature ->
1891212 """
190 url = f"{NWS_API_BASE}/alerts/active/area/{state}"
191 data = await make_nws_request(url)
192
193 if not data or "features" not in data:
194 return "Unable to fetch alerts or no alerts found."
195
196 if not data["features"]:
197 return "No active alerts for this state."
198
199 alerts = [format_alert(feature) for feature in data["features"]]
200 return "\n---\n".join(alerts)
201
202
203 @mcp.tool()
204 async def get_forecast(latitude: float, longitude: float) -> str:
205 """Get weather forecast for a location.
206
207 Args:
208 latitude: Latitude of the location
209 longitude: Longitude of the location
1213 Event: ${feature.properties.event}
1214 Area: ${feature.properties.areaDesc}
1215 Severity: ${feature.properties.severity}
1216 Status: ${feature.properties.status}
1217 Headline: ${feature.properties.headline}
1218 """.trimIndent()
1219 }
1220}
1221
1222// Extension function to fetch forecast information for given latitude and longitude
1223suspend fun HttpClient.getForecast(latitude: Double, longitude: Double): List<String> {
1224 val points = this.get("/points/$latitude,$longitude").body<PointsResponse>()
1225 val forecastUrl = points.properties.forecast ?: error("No forecast URL available")
1226 val forecast = this.get(forecastUrl).body<ForecastResponse>()
1227 return forecast.properties.periods.map { period ->
2101228 """
211 # First get the forecast grid endpoint
212 points_url = f"{NWS_API_BASE}/points/{latitude},{longitude}"
213 points_data = await make_nws_request(points_url)
214
215 if not points_data:
216 return "Unable to fetch forecast data for this location."
217
218 # Get the forecast URL from the points response
219 forecast_url = points_data["properties"]["forecast"]
220 forecast_data = await make_nws_request(forecast_url)
221
222 if not forecast_data:
223 return "Unable to fetch detailed forecast."
224
225 # Format the periods into a readable forecast
226 periods = forecast_data["properties"]["periods"]
227 forecasts = []
228 for period in periods[:5]: # Only show next 5 periods
229 forecast = f"""
230 {period["name"]}:
231 Temperature: {period["temperature"]}°{period["temperatureUnit"]}
232 Wind: {period["windSpeed"]} {period["windDirection"]}
233 Forecast: {period["detailedForecast"]}
234 """
235 forecasts.append(forecast)
236
237 return "\n---\n".join(forecasts)
238 ```
239
240 ### Running the server
241
242 Finally, let's initialize and run the server:
243
244 ```python theme={null}
245 def main():
246 # Initialize and run the server
247 mcp.run(transport="stdio")
248
249
250 if __name__ == "__main__":
251 main()
252 ```
253
254 Your server is complete! Run `uv run weather.py` to start the MCP server, which will listen for messages from MCP hosts.
255
256 Let's now test your server from an existing MCP host, Claude for Desktop.
257
258 ## Testing your server with Claude for Desktop
259
260 <Note>
261 Claude for Desktop is not yet available on Linux. Linux users can proceed to the [Building a client](/docs/2024-11-05/develop/build-client) tutorial to build an MCP client that connects to the server we just built.
262 </Note>
263
264 First, make sure you have Claude for Desktop installed. [You can install the latest version
265 here.](https://claude.ai/download) If you already have Claude for Desktop, **make sure it's updated to the latest version.**
266
267 We'll need to configure Claude for Desktop for whichever MCP servers you want to use. To do this, open your Claude for Desktop App configuration at `~/Library/Application Support/Claude/claude_desktop_config.json` in a text editor. Make sure to create the file if it doesn't exist.
268
269 For example, if you have [VS Code](https://code.visualstudio.com/) installed:
270
271 <CodeGroup>
272 ```bash macOS/Linux theme={null}
273 code ~/Library/Application\ Support/Claude/claude_desktop_config.json
274 ```
275
276 ```powershell Windows theme={null}
277 code $env:AppData\Claude\claude_desktop_config.json
278 ```
279 </CodeGroup>
280
281 You'll then add your servers in the `mcpServers` key. The MCP UI elements will only show up in Claude for Desktop if at least one server is properly configured.
282
283 In this case, we'll add our single weather server like so:
284
285 <CodeGroup>
286 ```json macOS/Linux theme={null}
287 {
288 "mcpServers": {
289 "weather": {
290 "command": "uv",
291 "args": [
292 "--directory",
293 "/ABSOLUTE/PATH/TO/PARENT/FOLDER/weather",
294 "run",
295 "weather.py"
296 ]
297 }
1229 ${period.name}:
1230 Temperature: ${period.temperature}°${period.temperatureUnit}
1231 Wind: ${period.windSpeed} ${period.windDirection}
1232 ${period.shortForecast}
1233 """.trimIndent()
1234 }
1235}
1236
1237@Serializable
1238data class PointsResponse(val properties: PointsProperties)
1239
1240@Serializable
1241data class PointsProperties(val forecast: String? = null)
1242
1243@Serializable
1244data class ForecastResponse(val properties: ForecastProperties)
1245
1246@Serializable
1247data class ForecastProperties(val periods: List<ForecastPeriod> = emptyList())
1248
1249@Serializable
1250data class ForecastPeriod(
1251 val name: String? = null,
1252 val temperature: Int? = null,
1253 val temperatureUnit: String? = null,
1254 val windSpeed: String? = null,
1255 val windDirection: String? = null,
1256 val shortForecast: String? = null,
1257)
1258
1259@Serializable
1260data class AlertsResponse(val features: List<AlertFeature> = emptyList())
1261
1262@Serializable
1263data class AlertFeature(val properties: AlertProperties)
1264
1265@Serializable
1266data class AlertProperties(
1267 val event: String? = null,
1268 val areaDesc: String? = null,
1269 val severity: String? = null,
1270 val status: String? = null,
1271 val headline: String? = null,
1272)
1273```
1274
1275### Implementing tool execution
1276
1277The tool execution handler is responsible for actually executing the logic of each tool. Let's add it:
1278
1279```kotlin theme={null}
1280// Register weather tools
1281
1282server.addTool(
1283 name = "get_alerts",
1284 description = "Get weather alerts for a US state. Input is a two-letter US state code (e.g. CA, NY)",
1285 inputSchema = ToolSchema(
1286 properties = buildJsonObject {
1287 putJsonObject("state") {
1288 put("type", "string")
1289 put("description", "Two-letter US state code (e.g. CA, NY)")
1290 }
1291 },
1292 required = listOf("state"),
1293 ),
1294) { request ->
1295 val state = request.arguments?.get("state")?.jsonPrimitive?.content
1296 ?: return@addTool CallToolResult(
1297 content = listOf(TextContent("The 'state' parameter is required.")),
1298 )
1299
1300 val alerts = httpClient.getAlerts(state)
1301 CallToolResult(content = alerts.map { TextContent(it) })
1302}
1303
1304server.addTool(
1305 name = "get_forecast",
1306 description = "Get weather forecast for a location. Note: only US locations are supported by the NWS API.",
1307 inputSchema = ToolSchema(
1308 properties = buildJsonObject {
1309 putJsonObject("latitude") {
1310 put("type", "number")
1311 put("description", "Latitude of the location")
1312 }
1313 putJsonObject("longitude") {
1314 put("type", "number")
1315 put("description", "Longitude of the location")
1316 }
1317 },
1318 required = listOf("latitude", "longitude"),
1319 ),
1320) { request ->
1321 val latitude = request.arguments?.get("latitude")?.jsonPrimitive?.doubleOrNull
1322 val longitude = request.arguments?.get("longitude")?.jsonPrimitive?.doubleOrNull
1323 if (latitude == null || longitude == null) {
1324 return@addTool CallToolResult(
1325 content = listOf(TextContent("The 'latitude' and 'longitude' parameters are required.")),
1326 )
1327 }
1328
1329 val forecast = httpClient.getForecast(latitude, longitude)
1330 CallToolResult(content = forecast.map { TextContent(it) })
1331}
1332```
1333
1334### Running the server
1335
1336Finally, implement the main function to run the server:
1337
1338```kotlin theme={null}
1339fun main() = runMcpServer()
1340```
1341
1342You can run the server directly during development:
1343
1344```bash theme={null}
1345./gradlew run
1346```
1347
1348For production use, build the shadow JAR:
1349
1350```bash theme={null}
1351./gradlew build
1352java -jar build/libs/weather-0.1.0-all.jar
1353```
1354
1355Let's now test your server from an existing MCP host, Claude for Desktop.
1356
1357## Testing your server with Claude for Desktop
1358
1359> [!NOTE]
1360>
1361> Claude for Desktop is not yet available on Linux. Linux users can proceed to the [Building a client](/docs/2024-11-05/develop/build-client) tutorial to build an MCP client that connects to the server we just built.
1362
1363First, make sure you have Claude for Desktop installed. [You can install the latest version
1364here.](https://claude.ai/download) If you already have Claude for Desktop, **make sure it's updated to the latest version.**
1365
1366We'll need to configure Claude for Desktop for whichever MCP servers you want to use.
1367To do this, open your Claude for Desktop App configuration at `~/Library/Application Support/Claude/claude_desktop_config.json` in a text editor.
1368Make sure to create the file if it doesn't exist.
1369
1370For example, if you have [VS Code](https://code.visualstudio.com/) installed:
1371
1372```bash macOS/Linux theme={null}
1373code ~/Library/Application\ Support/Claude/claude_desktop_config.json
1374```
1375
1376```powershell Windows theme={null}
1377code $env:AppData\Claude\claude_desktop_config.json
1378```
1379
1380You'll then add your servers in the `mcpServers` key.
1381The MCP UI elements will only show up in Claude for Desktop if at least one server is properly configured.
1382
1383In this case, we'll add our single weather server like so:
1384
1385```json macOS/Linux theme={null}
1386{
1387 "mcpServers": {
1388 "weather": {
1389 "command": "java",
1390 "args": [
1391 "-jar",
1392 "/ABSOLUTE/PATH/TO/PARENT/FOLDER/weather/build/libs/weather-0.1.0-all.jar"
1393 ]
1394 }
1395 }
1396}
1397```
1398
1399```json Windows theme={null}
1400{
1401 "mcpServers": {
1402 "weather": {
1403 "command": "java",
1404 "args": [
1405 "-jar",
1406 "C:\\PATH\\TO\\PARENT\\FOLDER\\weather\\build\\libs\\weather-0.1.0-all.jar"
1407 ]
1408 }
1409 }
1410}
1411```
1412
1413This tells Claude for Desktop:
1414
14151. There's an MCP server named "weather"
14162. Launch it by running `java -jar /ABSOLUTE/PATH/TO/PARENT/FOLDER/weather/build/libs/weather-0.1.0-all.jar`
1417
1418Save the file, and restart **Claude for Desktop**.
1419
1420**C#**
1421
1422Let's get started with building our weather server! [You can find the complete code for what we'll be building here.](https://github.com/modelcontextprotocol/csharp-sdk/tree/main/samples/QuickstartWeatherServer)
1423
1424### Prerequisite knowledge
1425
1426This quickstart assumes you have familiarity with:
1427
1428* C#
1429* LLMs like Claude
1430* .NET 8 or higher
1431
1432### Logging in MCP Servers
1433
1434When implementing MCP servers, be careful about how you handle logging:
1435
1436**For STDIO-based servers:** Never use `Console.WriteLine()` or `Console.Write()`, as they write to standard output (stdout). Writing to stdout will corrupt the JSON-RPC messages and break your server.
1437
1438**For HTTP-based servers:** Standard output logging is fine since it doesn't interfere with HTTP responses.
1439
1440### Best Practices
1441
1442* Use a logging library that writes to stderr or files.
1443
1444### System requirements
1445
1446* [.NET 8 SDK](https://dotnet.microsoft.com/download/dotnet/8.0) or higher installed.
1447
1448### Set up your environment
1449
1450First, let's install `dotnet` if you haven't already. You can download `dotnet` from [official Microsoft .NET website](https://dotnet.microsoft.com/download/). Verify your `dotnet` installation:
1451
1452```bash theme={null}
1453dotnet --version
1454```
1455
1456Now, let's create and set up your project:
1457
1458```bash macOS/Linux theme={null}
1459# Create a new directory for our project
1460mkdir weather
1461cd weather
1462# Initialize a new C# project
1463dotnet new console
1464```
1465
1466```powershell Windows theme={null}
1467# Create a new directory for our project
1468mkdir weather
1469cd weather
1470# Initialize a new C# project
1471dotnet new console
1472```
1473
1474After running `dotnet new console`, you will be presented with a new C# project.
1475You can open the project in your favorite IDE, such as [Visual Studio](https://visualstudio.microsoft.com/) or [Rider](https://www.jetbrains.com/rider/).
1476Alternatively, you can create a C# application using the [Visual Studio project wizard](https://learn.microsoft.com/en-us/visualstudio/get-started/csharp/tutorial-console?view=vs-2022).
1477After creating the project, add NuGet package for the Model Context Protocol SDK and hosting:
1478
1479```bash theme={null}
1480# Add the Model Context Protocol SDK NuGet package
1481dotnet add package ModelContextProtocol --prerelease
1482# Add the .NET Hosting NuGet package
1483dotnet add package Microsoft.Extensions.Hosting
1484```
1485
1486Now let’s dive into building your server.
1487
1488## Building your server
1489
1490Open the `Program.cs` file in your project and replace its contents with the following code:
1491
1492```csharp theme={null}
1493using Microsoft.Extensions.DependencyInjection;
1494using Microsoft.Extensions.Hosting;
1495using ModelContextProtocol;
1496using System.Net.Http.Headers;
1497
1498var builder = Host.CreateEmptyApplicationBuilder(settings: null);
1499
1500builder.Services.AddMcpServer()
1501 .WithStdioServerTransport()
1502 .WithToolsFromAssembly();
1503
1504builder.Services.AddSingleton(_ =>
1505{
1506 var client = new HttpClient() { BaseAddress = new Uri("https://api.weather.gov") };
1507 client.DefaultRequestHeaders.UserAgent.Add(new ProductInfoHeaderValue("weather-tool", "1.0"));
1508 return client;
1509});
1510
1511var app = builder.Build();
1512
1513await app.RunAsync();
1514```
1515
1516> [!NOTE]
1517>
1518> When creating the `ApplicationHostBuilder`, ensure you use `CreateEmptyApplicationBuilder` instead of `CreateDefaultBuilder`. This ensures that the server does not write any additional messages to the console. This is only necessary for servers using STDIO transport.
1519
1520This code sets up a basic console application that uses the Model Context Protocol SDK to create an MCP server with standard I/O transport.
1521
1522### Weather API helper functions
1523
1524Create an extension class for `HttpClient` which helps simplify JSON request handling:
1525
1526```csharp theme={null}
1527using System.Text.Json;
1528
1529internal static class HttpClientExt
1530{
1531 public static async Task<JsonDocument> ReadJsonDocumentAsync(this HttpClient client, string requestUri)
1532 {
1533 using var response = await client.GetAsync(requestUri);
1534 response.EnsureSuccessStatusCode();
1535 return await JsonDocument.ParseAsync(await response.Content.ReadAsStreamAsync());
1536 }
1537}
1538```
1539
1540Next, define a class with the tool execution handlers for querying and converting responses from the National Weather Service API:
1541
1542```csharp theme={null}
1543using ModelContextProtocol.Server;
1544using System.ComponentModel;
1545using System.Globalization;
1546using System.Text.Json;
1547
1548namespace QuickstartWeatherServer.Tools;
1549
1550[McpServerToolType]
1551public static class WeatherTools
1552{
1553 [McpServerTool, Description("Get weather alerts for a US state code.")]
1554 public static async Task<string> GetAlerts(
1555 HttpClient client,
1556 [Description("The US state code to get alerts for.")] string state)
1557 {
1558 using var jsonDocument = await client.ReadJsonDocumentAsync($"/alerts/active/area/{state}");
1559 var jsonElement = jsonDocument.RootElement;
1560 var alerts = jsonElement.GetProperty("features").EnumerateArray();
1561
1562 if (!alerts.Any())
1563 {
1564 return "No active alerts for this state.";
2981565 }
1566
1567 return string.Join("\n--\n", alerts.Select(alert =>
1568 {
1569 JsonElement properties = alert.GetProperty("properties");
1570 return $"""
1571 Event: {properties.GetProperty("event").GetString()}
1572 Area: {properties.GetProperty("areaDesc").GetString()}
1573 Severity: {properties.GetProperty("severity").GetString()}
1574 Description: {properties.GetProperty("description").GetString()}
1575 Instruction: {properties.GetProperty("instruction").GetString()}
1576 """;
1577 }));
1578 }
1579
1580 [McpServerTool, Description("Get weather forecast for a location.")]
1581 public static async Task<string> GetForecast(
1582 HttpClient client,
1583 [Description("Latitude of the location.")] double latitude,
1584 [Description("Longitude of the location.")] double longitude)
1585 {
1586 var pointUrl = string.Create(CultureInfo.InvariantCulture, $"/points/{latitude},{longitude}");
1587 using var jsonDocument = await client.ReadJsonDocumentAsync(pointUrl);
1588 var forecastUrl = jsonDocument.RootElement.GetProperty("properties").GetProperty("forecast").GetString()
1589 ?? throw new Exception($"No forecast URL provided by {client.BaseAddress}points/{latitude},{longitude}");
1590
1591 using var forecastDocument = await client.ReadJsonDocumentAsync(forecastUrl);
1592 var periods = forecastDocument.RootElement.GetProperty("properties").GetProperty("periods").EnumerateArray();
1593
1594 return string.Join("\n---\n", periods.Select(period => $"""
1595 {period.GetProperty("name").GetString()}
1596 Temperature: {period.GetProperty("temperature").GetInt32()}°F
1597 Wind: {period.GetProperty("windSpeed").GetString()} {period.GetProperty("windDirection").GetString()}
1598 Forecast: {period.GetProperty("detailedForecast").GetString()}
1599 """));
1600 }
1601}
1602```
1603
1604### Running the server
1605
1606Finally, run the server using the following command:
1607
1608```bash theme={null}
1609dotnet run
1610```
1611
1612This will start the server and listen for incoming requests on standard input/output.
1613
1614## Testing your server with Claude for Desktop
1615
1616> [!NOTE]
1617>
1618> Claude for Desktop is not yet available on Linux. Linux users can proceed to the [Building a client](/docs/2024-11-05/develop/build-client) tutorial to build an MCP client that connects to the server we just built.
1619
1620First, make sure you have Claude for Desktop installed. [You can install the latest version
1621here.](https://claude.ai/download) If you already have Claude for Desktop, **make sure it's updated to the latest version.**
1622We'll need to configure Claude for Desktop for whichever MCP servers you want to use. To do this, open your Claude for Desktop App configuration at `~/Library/Application Support/Claude/claude_desktop_config.json` in a text editor. Make sure to create the file if it doesn't exist.
1623For example, if you have [VS Code](https://code.visualstudio.com/) installed:
1624
1625```bash macOS/Linux theme={null}
1626code ~/Library/Application\ Support/Claude/claude_desktop_config.json
1627```
1628
1629```powershell Windows theme={null}
1630code $env:AppData\Claude\claude_desktop_config.json
1631```
1632
1633You'll then add your servers in the `mcpServers` key. The MCP UI elements will only show up in Claude for Desktop if at least one server is properly configured.
1634In this case, we'll add our single weather server like so:
1635
1636```json macOS/Linux theme={null}
1637{
1638 "mcpServers": {
1639 "weather": {
1640 "command": "dotnet",
1641 "args": ["run", "--project", "/ABSOLUTE/PATH/TO/PROJECT", "--no-build"]
1642 }
1643 }
1644}
1645```
1646
1647```json Windows theme={null}
1648{
1649 "mcpServers": {
1650 "weather": {
1651 "command": "dotnet",
1652 "args": [
1653 "run",
1654 "--project",
1655 "C:\\ABSOLUTE\\PATH\\TO\\PROJECT",
1656 "--no-build"
1657 ]
1658 }
1659 }
1660}
1661```
1662
1663This tells Claude for Desktop:
1664
16651. There's an MCP server named "weather"
16662. Launch it by running `dotnet run /ABSOLUTE/PATH/TO/PROJECT`
1667 Save the file, and restart **Claude for Desktop**.
1668
1669**Ruby**
1670
1671Let's get started with building our weather server! [You can find the complete code for what we'll be building here.](https://github.com/modelcontextprotocol/quickstart-resources/tree/main/weather-server-ruby)
1672
1673### Prerequisite knowledge
1674
1675This quickstart assumes you have familiarity with:
1676
1677* Ruby
1678* LLMs like Claude
1679
1680### Logging in MCP Servers
1681
1682When implementing MCP servers, be careful about how you handle logging:
1683
1684**For STDIO-based servers:** Never use `puts` or `print`, as they write to standard output (stdout) by default. Writing to stdout will corrupt the JSON-RPC messages and break your server.
1685
1686**For HTTP-based servers:** Standard output logging is fine since it doesn't interfere with HTTP responses.
1687
1688### Best Practices
1689
1690* Use a logging library that writes to stderr or files.
1691
1692### Quick Examples
1693
1694```ruby theme={null}
1695# ❌ Bad (STDIO)
1696puts "Processing request"
1697
1698# ✅ Good (STDIO)
1699require "logger"
1700logger = Logger.new($stderr)
1701logger.info("Processing request")
1702```
1703
1704### System requirements
1705
1706* Ruby 2.7 or higher installed.
1707
1708### Set up your environment
1709
1710First, let's make sure you have Ruby installed. You can check by running:
1711
1712```bash theme={null}
1713ruby --version
1714```
1715
1716Now, let's create and set up our project:
1717
1718```bash macOS/Linux theme={null}
1719# Create a new directory for our project
1720mkdir weather
1721cd weather
1722
1723# Create a Gemfile
1724bundle init
1725
1726# Add the MCP SDK dependency
1727bundle add mcp
1728
1729# Create our server file
1730touch weather.rb
1731```
1732
1733```powershell Windows theme={null}
1734# Create a new directory for our project
1735mkdir weather
1736cd weather
1737
1738# Create a Gemfile
1739bundle init
1740
1741# Add the MCP SDK dependency
1742bundle add mcp
1743
1744# Create our server file
1745new-item weather.rb
1746```
1747
1748Now let's dive into building your server.
1749
1750## Building your server
1751
1752### Importing packages and setting up constants
1753
1754Open `weather.rb` and add these requires and constants at the top:
1755
1756```ruby theme={null}
1757require "json"
1758require "mcp"
1759require "net/http"
1760require "uri"
1761
1762NWS_API_BASE = "https://api.weather.gov"
1763USER_AGENT = "weather-app/1.0"
1764```
1765
1766The `mcp` gem provides the Model Context Protocol SDK for Ruby, with classes for server implementation and stdio transport.
1767
1768### Helper methods
1769
1770Next, let's add helper methods for querying and formatting data from the National Weather Service API:
1771
1772```ruby theme={null}
1773module HelperMethods
1774 def make_nws_request(url)
1775 uri = URI(url)
1776 request = Net::HTTP::Get.new(uri)
1777 request["User-Agent"] = USER_AGENT
1778 request["Accept"] = "application/geo+json"
1779
1780 response = Net::HTTP.start(uri.hostname, uri.port, use_ssl: true) do |http|
1781 http.request(request)
1782 end
1783
1784 raise "HTTP #{response.code}: #{response.message}" unless response.is_a?(Net::HTTPSuccess)
1785
1786 JSON.parse(response.body)
1787 end
1788
1789 def format_alert(feature)
1790 properties = feature["properties"]
1791
1792 <<~ALERT
1793 Event: #{properties["event"] || "Unknown"}
1794 Area: #{properties["areaDesc"] || "Unknown"}
1795 Severity: #{properties["severity"] || "Unknown"}
1796 Description: #{properties["description"] || "No description available"}
1797 Instructions: #{properties["instruction"] || "No specific instructions provided"}
1798 ALERT
1799 end
1800end
1801```
1802
1803### Implementing tool execution
1804
1805Now let's define our tool classes. Each tool subclasses `MCP::Tool` and implements the tool logic:
1806
1807```ruby theme={null}
1808class GetAlerts < MCP::Tool
1809 extend HelperMethods
1810
1811 tool_name "get_alerts"
1812 description "Get weather alerts for a US state"
1813 input_schema(
1814 properties: {
1815 state: {
1816 type: "string",
1817 description: "Two-letter US state code (e.g. CA, NY)"
2991818 }
300 ```
301
302 ```json Windows theme={null}
303 {
304 "mcpServers": {
305 "weather": {
306 "command": "uv",
307 "args": [
308 "--directory",
309 "C:\\ABSOLUTE\\PATH\\TO\\PARENT\\FOLDER\\weather",
310 "run",
311 "weather.py"
312 ]
313 }
1819 },
1820 required: ["state"]
1821 )
1822
1823 def self.call(state:)
1824 url = "#{NWS_API_BASE}/alerts/active/area/#{state.upcase}"
1825 data = make_nws_request(url)
1826
1827 if data["features"].empty?
1828 return MCP::Tool::Response.new([{
1829 type: "text",
1830 text: "No active alerts for this state."
1831 }])
1832 end
1833
1834 alerts = data["features"].map { |feature| format_alert(feature) }
1835 MCP::Tool::Response.new([{
1836 type: "text",
1837 text: alerts.join("\n---\n")
1838 }])
1839 end
1840end
1841
1842class GetForecast < MCP::Tool
1843 extend HelperMethods
1844
1845 tool_name "get_forecast"
1846 description "Get weather forecast for a location"
1847 input_schema(
1848 properties: {
1849 latitude: {
1850 type: "number",
1851 description: "Latitude of the location"
1852 },
1853 longitude: {
1854 type: "number",
1855 description: "Longitude of the location"
1856 }
1857 },
1858 required: ["latitude", "longitude"]
1859 )
1860
1861 def self.call(latitude:, longitude:)
1862 # First get the forecast grid endpoint.
1863 points_url = "#{NWS_API_BASE}/points/#{latitude},#{longitude}"
1864 points_data = make_nws_request(points_url)
1865
1866 # Get the forecast URL from the points response.
1867 forecast_url = points_data["properties"]["forecast"]
1868 forecast_data = make_nws_request(forecast_url)
1869
1870 # Format the periods into a readable forecast.
1871 periods = forecast_data["properties"]["periods"]
1872 forecasts = periods.first(5).map do |period|
1873 <<~FORECAST
1874 #{period["name"]}:
1875 Temperature: #{period["temperature"]}°#{period["temperatureUnit"]}
1876 Wind: #{period["windSpeed"]} #{period["windDirection"]}
1877 Forecast: #{period["detailedForecast"]}
1878 FORECAST
1879 end
1880
1881 MCP::Tool::Response.new([{
1882 type: "text",
1883 text: forecasts.join("\n---\n")
1884 }])
1885 end
1886end
1887```
1888
1889### Running the server
1890
1891Finally, initialize and run the server:
1892
1893```ruby theme={null}
1894server = MCP::Server.new(
1895 name: "weather",
1896 version: "1.0.0",
1897 tools: [GetAlerts, GetForecast]
1898)
1899
1900transport = MCP::Server::Transports::StdioTransport.new(server)
1901transport.open
1902```
1903
1904Your server is complete! Run `bundle exec ruby weather.rb` to start the MCP server, which will listen for messages from MCP hosts.
1905
1906Let's now test your server from an existing MCP host, Claude for Desktop.
1907
1908## Testing your server with Claude for Desktop
1909
1910> [!NOTE]
1911>
1912> Claude for Desktop is not yet available on Linux. Linux users can proceed to the [Building a client](/docs/2024-11-05/develop/build-client) tutorial to build an MCP client that connects to the server we just built.
1913
1914First, make sure you have Claude for Desktop installed. [You can install the latest version here.](https://claude.ai/download) If you already have Claude for Desktop, **make sure it's updated to the latest version.**
1915
1916We'll need to configure Claude for Desktop for whichever MCP servers you want to use. To do this, open your Claude for Desktop App configuration at `~/Library/Application Support/Claude/claude_desktop_config.json` in a text editor. Make sure to create the file if it doesn't exist.
1917
1918For example, if you have [VS Code](https://code.visualstudio.com/) installed:
1919
1920```bash macOS/Linux theme={null}
1921code ~/Library/Application\ Support/Claude/claude_desktop_config.json
1922```
1923
1924```powershell Windows theme={null}
1925code $env:AppData\Claude\claude_desktop_config.json
1926```
1927
1928You'll then add your servers in the `mcpServers` key. The MCP UI elements will only show up in Claude for Desktop if at least one server is properly configured.
1929
1930In this case, we'll add our single weather server like so:
1931
1932```json macOS/Linux theme={null}
1933{
1934 "mcpServers": {
1935 "weather": {
1936 "command": "bundle",
1937 "args": ["exec", "ruby", "weather.rb"],
1938 "cwd": "/ABSOLUTE/PATH/TO/PARENT/FOLDER/weather"
1939 }
1940 }
1941}
1942```
1943
1944```json Windows theme={null}
1945{
1946 "mcpServers": {
1947 "weather": {
1948 "command": "bundle",
1949 "args": ["exec", "ruby", "weather.rb"],
1950 "cwd": "C:\\ABSOLUTE\\PATH\\TO\\PARENT\\FOLDER\\weather"
1951 }
1952 }
1953}
1954```
1955
1956> [!NOTE]
1957>
1958> Make sure you pass in the absolute path to your project directory in the `cwd` field. You can get this by running `pwd` on macOS/Linux or `cd` on Windows Command Prompt from your project directory. On Windows, remember to use double backslashes (`\\`) or forward slashes (`/`) in the JSON path.
1959
1960This tells Claude for Desktop:
1961
19621. There's an MCP server named "weather"
19632. Launch it by running `bundle exec ruby weather.rb` in the specified directory
1964
1965Save the file, and restart **Claude for Desktop**.
1966
1967**Rust**
1968
1969Let's get started with building our weather server! [You can find the complete code for what we'll be building here.](https://github.com/modelcontextprotocol/quickstart-resources/tree/main/weather-server-rust)
1970
1971### Prerequisite knowledge
1972
1973This quickstart assumes you have familiarity with:
1974
1975* Rust programming language
1976* Async/await in Rust
1977* LLMs like Claude
1978
1979### Logging in MCP Servers
1980
1981When implementing MCP servers, be careful about how you handle logging:
1982
1983**For STDIO-based servers:** Never use `println!()` or `print!()`, as they write to standard output (stdout). Writing to stdout will corrupt the JSON-RPC messages and break your server.
1984
1985**For HTTP-based servers:** Standard output logging is fine since it doesn't interfere with HTTP responses.
1986
1987### Best Practices
1988
1989* Use a logging library that writes to stderr or files, such as `tracing` or `log` in Rust.
1990* Configure your logging framework to avoid stdout output.
1991
1992### Quick Examples
1993
1994```rust theme={null}
1995// ❌ Bad (STDIO)
1996println!("Processing request");
1997
1998// ✅ Good (STDIO)
1999eprintln!("Processing request"); // writes to stderr
2000```
2001
2002### System requirements
2003
2004* Rust 1.70 or higher installed.
2005* Cargo (comes with Rust installation).
2006
2007### Set up your environment
2008
2009First, let's install Rust if you haven't already. You can install Rust from [rust-lang.org](https://www.rust-lang.org/tools/install):
2010
2011```bash macOS/Linux theme={null}
2012curl --proto '=https' --tlsv1.2 -sSf https://sh.rustup.rs | sh
2013```
2014
2015```powershell Windows theme={null}
2016# Download and run rustup-init.exe from https://rustup.rs/
2017```
2018
2019Verify your Rust installation:
2020
2021```bash theme={null}
2022rustc --version
2023cargo --version
2024```
2025
2026Now, let's create and set up our project:
2027
2028```bash macOS/Linux theme={null}
2029# Create a new Rust project
2030cargo new weather
2031cd weather
2032```
2033
2034```powershell Windows theme={null}
2035# Create a new Rust project
2036cargo new weather
2037cd weather
2038```
2039
2040Update your `Cargo.toml` to add the required dependencies:
2041
2042```toml Cargo.toml theme={null}
2043[package]
2044name = "weather"
2045version = "0.1.0"
2046edition = "2024"
2047
2048[dependencies]
2049rmcp = { version = "0.3", features = ["server", "macros", "transport-io"] }
2050tokio = { version = "1.46", features = ["full"] }
2051reqwest = { version = "0.12", features = ["json"] }
2052serde = { version = "1.0", features = ["derive"] }
2053serde_json = "1.0"
2054anyhow = "1.0"
2055tracing = "0.1"
2056tracing-subscriber = { version = "0.3", features = ["env-filter", "std", "fmt"] }
2057```
2058
2059Now let's dive into building your server.
2060
2061## Building your server
2062
2063### Importing packages and constants
2064
2065Open `src/main.rs` and add these imports and constants at the top:
2066
2067```rust theme={null}
2068use anyhow::Result;
2069use rmcp::{
2070 ServerHandler, ServiceExt,
2071 handler::server::{router::tool::ToolRouter, tool::Parameters},
2072 model::*,
2073 schemars, tool, tool_handler, tool_router,
2074};
2075use serde::Deserialize;
2076use serde::de::DeserializeOwned;
2077
2078const NWS_API_BASE: &str = "https://api.weather.gov";
2079const USER_AGENT: &str = "weather-app/1.0";
2080```
2081
2082The `rmcp` crate provides the Model Context Protocol SDK for Rust, with features for server implementation, procedural macros, and stdio transport.
2083
2084### Data structures
2085
2086Next, let's define the data structures for deserializing responses from the National Weather Service API:
2087
2088```rust theme={null}
2089#[derive(Debug, Deserialize)]
2090struct AlertsResponse {
2091 features: Vec<AlertFeature>,
2092}
2093
2094#[derive(Debug, Deserialize)]
2095struct AlertFeature {
2096 properties: AlertProperties,
2097}
2098
2099#[derive(Debug, Deserialize)]
2100struct AlertProperties {
2101 event: Option<String>,
2102 #[serde(rename = "areaDesc")]
2103 area_desc: Option<String>,
2104 severity: Option<String>,
2105 description: Option<String>,
2106 instruction: Option<String>,
2107}
2108
2109#[derive(Debug, Deserialize)]
2110struct PointsResponse {
2111 properties: PointsProperties,
2112}
2113
2114#[derive(Debug, Deserialize)]
2115struct PointsProperties {
2116 forecast: String,
2117}
2118
2119#[derive(Debug, Deserialize)]
2120struct ForecastResponse {
2121 properties: ForecastProperties,
2122}
2123
2124#[derive(Debug, Deserialize)]
2125struct ForecastProperties {
2126 periods: Vec<ForecastPeriod>,
2127}
2128
2129#[derive(Debug, Deserialize)]
2130struct ForecastPeriod {
2131 name: String,
2132 temperature: i32,
2133 #[serde(rename = "temperatureUnit")]
2134 temperature_unit: String,
2135 #[serde(rename = "windSpeed")]
2136 wind_speed: String,
2137 #[serde(rename = "windDirection")]
2138 wind_direction: String,
2139 #[serde(rename = "detailedForecast")]
2140 detailed_forecast: String,
2141}
2142```
2143
2144Now define the request types that MCP clients will send:
2145
2146```rust theme={null}
2147#[derive(serde::Deserialize, schemars::JsonSchema)]
2148pub struct MCPForecastRequest {
2149 latitude: f32,
2150 longitude: f32,
2151}
2152
2153#[derive(serde::Deserialize, schemars::JsonSchema)]
2154pub struct MCPAlertRequest {
2155 state: String,
2156}
2157```
2158
2159### Helper functions
2160
2161Add helper functions for making API requests and formatting responses:
2162
2163```rust theme={null}
2164async fn make_nws_request<T: DeserializeOwned>(url: &str) -> Result<T> {
2165 let client = reqwest::Client::new();
2166 let rsp = client
2167 .get(url)
2168 .header(reqwest::header::USER_AGENT, USER_AGENT)
2169 .header(reqwest::header::ACCEPT, "application/geo+json")
2170 .send()
2171 .await?
2172 .error_for_status()?;
2173 Ok(rsp.json::<T>().await?)
2174}
2175
2176fn format_alert(feature: &AlertFeature) -> String {
2177 let props = &feature.properties;
2178 format!(
2179 "Event: {}\nArea: {}\nSeverity: {}\nDescription: {}\nInstructions: {}",
2180 props.event.as_deref().unwrap_or("Unknown"),
2181 props.area_desc.as_deref().unwrap_or("Unknown"),
2182 props.severity.as_deref().unwrap_or("Unknown"),
2183 props
2184 .description
2185 .as_deref()
2186 .unwrap_or("No description available"),
2187 props
2188 .instruction
2189 .as_deref()
2190 .unwrap_or("No specific instructions provided")
2191 )
2192}
2193
2194fn format_period(period: &ForecastPeriod) -> String {
2195 format!(
2196 "{}:\nTemperature: {}°{}\nWind: {} {}\nForecast: {}",
2197 period.name,
2198 period.temperature,
2199 period.temperature_unit,
2200 period.wind_speed,
2201 period.wind_direction,
2202 period.detailed_forecast
2203 )
2204}
2205```
2206
2207### Implementing the Weather server and tools
2208
2209Now let's implement the main Weather server struct with the tool handlers:
2210
2211```rust theme={null}
2212pub struct Weather {
2213 tool_router: ToolRouter<Weather>,
2214}
2215
2216#[tool_router]
2217impl Weather {
2218 fn new() -> Self {
2219 Self {
2220 tool_router: Self::tool_router(),
3142221 }
315 }
316 ```
317 </CodeGroup>
318
319 <Warning>
320 You may need to put the full path to the `uv` executable in the `command` field. You can get this by running `which uv` on macOS/Linux or `where uv` on Windows.
321 </Warning>
322
323 <Note>
324 Make sure you pass in the absolute path to your server. You can get this by running `pwd` on macOS/Linux or `cd` on Windows Command Prompt. On Windows, remember to use double backslashes (`\\`) or forward slashes (`/`) in the JSON path.
325 </Note>
326
327 This tells Claude for Desktop:
328
329 1. There's an MCP server named "weather"
330 2. To launch it by running `uv --directory /ABSOLUTE/PATH/TO/PARENT/FOLDER/weather run weather.py`
331
332 Save the file, and restart **Claude for Desktop**.
333 </Tab>
334
335 <Tab title="TypeScript">
336 Let's get started with building our weather server! [You can find the complete code for what we'll be building here.](https://github.com/modelcontextprotocol/quickstart-resources/tree/main/weather-server-typescript)
337
338 ### Prerequisite knowledge
339
340 This quickstart assumes you have familiarity with:
341
342 * TypeScript
343 * LLMs like Claude
344
345 ### Logging in MCP Servers
346
347 When implementing MCP servers, be careful about how you handle logging:
348
349 **For STDIO-based servers:** Never use `console.log()`, as it writes to standard output (stdout) by default. Writing to stdout will corrupt the JSON-RPC messages and break your server.
350
351 **For HTTP-based servers:** Standard output logging is fine since it doesn't interfere with HTTP responses.
352
353 ### Best Practices
354
355 * Use `console.error()` which writes to stderr, or use a logging library that writes to stderr or files.
356
357 ### Quick Examples
358
359 ```javascript theme={null}
360 // ❌ Bad (STDIO)
361 console.log("Server started");
362
363 // ✅ Good (STDIO)
364 console.error("Server started"); // stderr is safe
365 ```
366
367 ### System requirements
368
369 For TypeScript, make sure you have the latest version of Node installed.
370
371 ### Set up your environment
372
373 First, let's install Node.js and npm if you haven't already. You can download them from [nodejs.org](https://nodejs.org/).
374 Verify your Node.js installation:
375
376 ```bash theme={null}
377 node --version
378 npm --version
379 ```
380
381 For this tutorial, you'll need Node.js version 16 or higher.
382
383 Now, let's create and set up our project:
384
385 <CodeGroup>
386 ```bash macOS/Linux theme={null}
387 # Create a new directory for our project
388 mkdir weather
389 cd weather
390
391 # Initialize a new npm project
392 npm init -y
393
394 # Install dependencies
395 npm install @modelcontextprotocol/sdk zod@3
396 npm install -D @types/node typescript
397
398 # Create our files
399 mkdir src
400 touch src/index.ts
401 ```
402
403 ```powershell Windows theme={null}
404 # Create a new directory for our project
405 md weather
406 cd weather
407
408 # Initialize a new npm project
409 npm init -y
410
411 # Install dependencies
412 npm install @modelcontextprotocol/sdk zod@3
413 npm install -D @types/node typescript
414
415 # Create our files
416 md src
417 new-item src\index.ts
418 ```
419 </CodeGroup>
420
421 Update your package.json to add type: "module" and a build script:
422
423 ```json package.json theme={null}
424 {
425 "type": "module",
426 "bin": {
427 "weather": "./build/index.js"
428 },
429 "scripts": {
430 "build": "tsc && chmod 755 build/index.js"
431 },
432 "files": ["build"]
433 }
434 ```
435
436 Create a `tsconfig.json` in the root of your project:
437
438 ```json tsconfig.json theme={null}
439 {
440 "compilerOptions": {
441 "target": "ES2022",
442 "module": "Node16",
443 "moduleResolution": "Node16",
444 "outDir": "./build",
445 "rootDir": "./src",
446 "strict": true,
447 "esModuleInterop": true,
448 "skipLibCheck": true,
449 "forceConsistentCasingInFileNames": true
450 },
451 "include": ["src/**/*"],
452 "exclude": ["node_modules"]
453 }
454 ```
455
456 Now let's dive into building your server.
457
458 ## Building your server
459
460 ### Importing packages and setting up the instance
461
462 Add these to the top of your `src/index.ts`:
463
464 ```typescript theme={null}
465 import { McpServer } from "@modelcontextprotocol/sdk/server/mcp.js";
466 import { StdioServerTransport } from "@modelcontextprotocol/sdk/server/stdio.js";
467 import { z } from "zod";
468
469 const NWS_API_BASE = "https://api.weather.gov";
470 const USER_AGENT = "weather-app/1.0";
471
472 // Create server instance
473 const server = new McpServer({
474 name: "weather",
475 version: "1.0.0",
476 });
477 ```
478
479 ### Helper functions
480
481 Next, let's add our helper functions for querying and formatting the data from the National Weather Service API:
482
483 ```typescript theme={null}
484 // Helper function for making NWS API requests
485 async function makeNWSRequest<T>(url: string): Promise<T | null> {
486 const headers = {
487 "User-Agent": USER_AGENT,
488 Accept: "application/geo+json",
489 };
490
491 try {
492 const response = await fetch(url, { headers });
493 if (!response.ok) {
494 throw new Error(`HTTP error! status: ${response.status}`);
2222 }
2223
2224 #[tool(description = "Get weather alerts for a US state.")]
2225 async fn get_alerts(
2226 &self,
2227 Parameters(MCPAlertRequest { state }): Parameters<MCPAlertRequest>,
2228 ) -> String {
2229 let url = format!(
2230 "{}/alerts/active/area/{}",
2231 NWS_API_BASE,
2232 state.to_uppercase()
2233 );
2234
2235 match make_nws_request::<AlertsResponse>(&url).await {
2236 Ok(data) => {
2237 if data.features.is_empty() {
2238 "No active alerts for this state.".to_string()
2239 } else {
2240 data.features
2241 .iter()
2242 .map(format_alert)
2243 .collect::<Vec<_>>()
2244 .join("\n---\n")
2245 }
2246 }
2247 Err(_) => "Unable to fetch alerts or no alerts found.".to_string(),
4952248 }
496 return (await response.json()) as T;
497 } catch (error) {
498 console.error("Error making NWS request:", error);
499 return null;
500 }
501 }
502
503 interface AlertFeature {
504 properties: {
505 event?: string;
506 areaDesc?: string;
507 severity?: string;
508 status?: string;
509 headline?: string;
510 };
511 }
512
513 // Format alert data
514 function formatAlert(feature: AlertFeature): string {
515 const props = feature.properties;
516 return [
517 `Event: ${props.event || "Unknown"}`,
518 `Area: ${props.areaDesc || "Unknown"}`,
519 `Severity: ${props.severity || "Unknown"}`,
520 `Status: ${props.status || "Unknown"}`,
521 `Headline: ${props.headline || "No headline"}`,
522 "---",
523 ].join("\n");
524 }
525
526 interface ForecastPeriod {
527 name?: string;
528 temperature?: number;
529 temperatureUnit?: string;
530 windSpeed?: string;
531 windDirection?: string;
532 shortForecast?: string;
533 }
534
535 interface AlertsResponse {
536 features: AlertFeature[];
537 }
538
539 interface PointsResponse {
540 properties: {
541 forecast?: string;
542 };
543 }
544
545 interface ForecastResponse {
546 properties: {
547 periods: ForecastPeriod[];
548 };
549 }
550 ```
551
552 ### Implementing tool execution
553
554 The tool execution handler is responsible for actually executing the logic of each tool. Let's add it:
555
556 ```typescript theme={null}
557 // Register weather tools
558
559 server.registerTool(
560 "get_alerts",
561 {
562 description: "Get weather alerts for a state",
563 inputSchema: {
564 state: z
565 .string()
566 .length(2)
567 .describe("Two-letter state code (e.g. CA, NY)"),
568 },
569 },
570 async ({ state }) => {
571 const stateCode = state.toUpperCase();
572 const alertsUrl = `${NWS_API_BASE}/alerts?area=${stateCode}`;
573 const alertsData = await makeNWSRequest<AlertsResponse>(alertsUrl);
574
575 if (!alertsData) {
576 return {
577 content: [
578 {
579 type: "text",
580 text: "Failed to retrieve alerts data",
581 },
582 ],
583 };
2249 }
2250
2251 #[tool(description = "Get weather forecast for a location.")]
2252 async fn get_forecast(
2253 &self,
2254 Parameters(MCPForecastRequest {
2255 latitude,
2256 longitude,
2257 }): Parameters<MCPForecastRequest>,
2258 ) -> String {
2259 let points_url = format!("{NWS_API_BASE}/points/{latitude},{longitude}");
2260 let Ok(points_data) = make_nws_request::<PointsResponse>(&points_url).await else {
2261 return "Unable to fetch forecast data for this location.".to_string();
2262 };
2263
2264 let forecast_url = points_data.properties.forecast;
2265
2266 let Ok(forecast_data) = make_nws_request::<ForecastResponse>(&forecast_url).await else {
2267 return "Unable to fetch forecast data for this location.".to_string();
2268 };
2269
2270 let periods = &forecast_data.properties.periods;
2271 let forecast_summary: String = periods
2272 .iter()
2273 .take(5) // Next 5 periods only
2274 .map(format_period)
2275 .collect::<Vec<String>>()
2276 .join("\n---\n");
2277 forecast_summary
2278 }
2279}
2280```
2281
2282The `#[tool_router]` macro automatically generates the routing logic, and the `#[tool]` attribute marks methods as MCP tools.
2283
2284### Implementing the ServerHandler
2285
2286Implement the `ServerHandler` trait to define server capabilities:
2287
2288```rust theme={null}
2289#[tool_handler]
2290impl ServerHandler for Weather {
2291 fn get_info(&self) -> ServerInfo {
2292 ServerInfo {
2293 capabilities: ServerCapabilities::builder().enable_tools().build(),
2294 ..Default::default()
5842295 }
585
586 const features = alertsData.features || [];
587 if (!features.length) {
588 return {
589 content: [
590 {
591 type: "text",
592 text: `No active alerts for ${stateCode}`,
593 },
594 ],
595 };
596 }
597
598 const formattedAlerts = features.map(formatAlert);
599 const alertsText = `Active alerts for ${stateCode}:\n\n${formattedAlerts.join("\n")}`;
600
601 return {
602 content: [
603 {
604 type: "text",
605 text: alertsText,
606 },
607 ],
608 };
609 },
610 );
611
612 server.registerTool(
613 "get_forecast",
614 {
615 description: "Get weather forecast for a location",
616 inputSchema: {
617 latitude: z
618 .number()
619 .min(-90)
620 .max(90)
621 .describe("Latitude of the location"),
622 longitude: z
623 .number()
624 .min(-180)
625 .max(180)
626 .describe("Longitude of the location"),
627 },
628 },
629 async ({ latitude, longitude }) => {
630 // Get grid point data
631 const pointsUrl = `${NWS_API_BASE}/points/${latitude.toFixed(4)},${longitude.toFixed(4)}`;
632 const pointsData = await makeNWSRequest<PointsResponse>(pointsUrl);
633
634 if (!pointsData) {
635 return {
636 content: [
637 {
638 type: "text",
639 text: `Failed to retrieve grid point data for coordinates: ${latitude}, ${longitude}. This location may not be supported by the NWS API (only US locations are supported).`,
640 },
641 ],
642 };
643 }
644
645 const forecastUrl = pointsData.properties?.forecast;
646 if (!forecastUrl) {
647 return {
648 content: [
649 {
650 type: "text",
651 text: "Failed to get forecast URL from grid point data",
652 },
653 ],
654 };
655 }
656
657 // Get forecast data
658 const forecastData = await makeNWSRequest<ForecastResponse>(forecastUrl);
659 if (!forecastData) {
660 return {
661 content: [
662 {
663 type: "text",
664 text: "Failed to retrieve forecast data",
665 },
666 ],
667 };
668 }
669
670 const periods = forecastData.properties?.periods || [];
671 if (periods.length === 0) {
672 return {
673 content: [
674 {
675 type: "text",
676 text: "No forecast periods available",
677 },
678 ],
679 };
680 }
681
682 // Format forecast periods
683 const formattedForecast = periods.map((period: ForecastPeriod) =>
684 [
685 `${period.name || "Unknown"}:`,
686 `Temperature: ${period.temperature || "Unknown"}°${period.temperatureUnit || "F"}`,
687 `Wind: ${period.windSpeed || "Unknown"} ${period.windDirection || ""}`,
688 `${period.shortForecast || "No forecast available"}`,
689 "---",
690 ].join("\n"),
691 );
692
693 const forecastText = `Forecast for ${latitude}, ${longitude}:\n\n${formattedForecast.join("\n")}`;
694
695 return {
696 content: [
697 {
698 type: "text",
699 text: forecastText,
700 },
701 ],
702 };
703 },
704 );
705 ```
706
707 ### Running the server
708
709 Finally, implement the main function to run the server:
710
711 ```typescript theme={null}
712 async function main() {
713 const transport = new StdioServerTransport();
714 await server.connect(transport);
715 console.error("Weather MCP Server running on stdio");
716 }
717
718 main().catch((error) => {
719 console.error("Fatal error in main():", error);
720 process.exit(1);
721 });
722 ```
723
724 Make sure to run `npm run build` to build your server! This is a very important step in getting your server to connect.
725
726 Let's now test your server from an existing MCP host, Claude for Desktop.
727
728 ## Testing your server with Claude for Desktop
729
730 <Note>
731 Claude for Desktop is not yet available on Linux. Linux users can proceed to the [Building a client](/docs/2024-11-05/develop/build-client) tutorial to build an MCP client that connects to the server we just built.
732 </Note>
733
734 First, make sure you have Claude for Desktop installed. [You can install the latest version
735 here.](https://claude.ai/download) If you already have Claude for Desktop, **make sure it's updated to the latest version.**
736
737 We'll need to configure Claude for Desktop for whichever MCP servers you want to use. To do this, open your Claude for Desktop App configuration at `~/Library/Application Support/Claude/claude_desktop_config.json` in a text editor. Make sure to create the file if it doesn't exist.
738
739 For example, if you have [VS Code](https://code.visualstudio.com/) installed:
740
741 <CodeGroup>
742 ```bash macOS/Linux theme={null}
743 code ~/Library/Application\ Support/Claude/claude_desktop_config.json
744 ```
745
746 ```powershell Windows theme={null}
747 code $env:AppData\Claude\claude_desktop_config.json
748 ```
749 </CodeGroup>
750
751 You'll then add your servers in the `mcpServers` key. The MCP UI elements will only show up in Claude for Desktop if at least one server is properly configured.
752
753 In this case, we'll add our single weather server like so:
754
755 <CodeGroup>
756 ```json macOS/Linux theme={null}
757 {
758 "mcpServers": {
759 "weather": {
760 "command": "node",
761 "args": ["/ABSOLUTE/PATH/TO/PARENT/FOLDER/weather/build/index.js"]
762 }
763 }
764 }
765 ```
766
767 ```json Windows theme={null}
768 {
769 "mcpServers": {
770 "weather": {
771 "command": "node",
772 "args": ["C:\\PATH\\TO\\PARENT\\FOLDER\\weather\\build\\index.js"]
773 }
774 }
775 }
776 ```
777 </CodeGroup>
778
779 This tells Claude for Desktop:
780
781 1. There's an MCP server named "weather"
782 2. Launch it by running `node /ABSOLUTE/PATH/TO/PARENT/FOLDER/weather/build/index.js`
783
784 Save the file, and restart **Claude for Desktop**.
785 </Tab>
786
787 <Tab title="Java">
788 <Note>
789 This is a quickstart demo based on Spring AI MCP auto-configuration and boot starters.
790 To learn how to create sync and async MCP Servers, manually, consult the [Java SDK Server](https://java.sdk.modelcontextprotocol.io/) documentation.
791 </Note>
792
793 Let's get started with building our weather server!
794 [You can find the complete code for what we'll be building here.](https://github.com/spring-projects/spring-ai-examples/tree/main/model-context-protocol/weather/starter-stdio-server)
795
796 For more information, see the [MCP Server Boot Starter](https://docs.spring.io/spring-ai/reference/api/mcp/mcp-server-boot-starter-docs.html) reference documentation.
797 For manual MCP Server implementation, refer to the [MCP Server Java SDK documentation](https://java.sdk.modelcontextprotocol.io/).
798
799 ### Logging in MCP Servers
800
801 When implementing MCP servers, be careful about how you handle logging:
802
803 **For STDIO-based servers:** Never use `System.out.println()` or `System.out.print()`, as they write to standard output (stdout). Writing to stdout will corrupt the JSON-RPC messages and break your server.
804
805 **For HTTP-based servers:** Standard output logging is fine since it doesn't interfere with HTTP responses.
806
807 ### Best Practices
808
809 * Use a logging library that writes to stderr or files.
810 * Ensure any configured logging library will not write to stdout.
811
812 ### System requirements
813
814 * Java 17 or higher installed.
815 * [Spring Boot 3.3.x](https://docs.spring.io/spring-boot/installing.html) or higher
816
817 ### Set up your environment
818
819 Use the [Spring Initializer](https://start.spring.io/) to bootstrap the project.
820
821 You will need to add the following dependencies:
822
823 <CodeGroup>
824 ```xml Maven theme={null}
825 <dependencies>
826 <dependency>
827 <groupId>org.springframework.ai</groupId>
828 <artifactId>spring-ai-starter-mcp-server</artifactId>
829 </dependency>
830
831 <dependency>
832 <groupId>org.springframework</groupId>
833 <artifactId>spring-web</artifactId>
834 </dependency>
835 </dependencies>
836 ```
837
838 ```groovy Gradle theme={null}
839 dependencies {
840 implementation platform("org.springframework.ai:spring-ai-starter-mcp-server")
841 implementation platform("org.springframework:spring-web")
842 }
843 ```
844 </CodeGroup>
845
846 Then configure your application by setting the application properties:
847
848 <CodeGroup>
849 ```bash application.properties theme={null}
850 spring.main.bannerMode=off
851 logging.pattern.console=
852 ```
853
854 ```yaml application.yml theme={null}
855 logging:
856 pattern:
857 console:
858 spring:
859 main:
860 banner-mode: off
861 ```
862 </CodeGroup>
863
864 The [Server Configuration Properties](https://docs.spring.io/spring-ai/reference/api/mcp/mcp-server-boot-starter-docs.html#_configuration_properties) documents all available properties.
865
866 Now let's dive into building your server.
867
868 ## Building your server
869
870 ### Weather Service
871
872 Let's implement a [WeatherService.java](https://github.com/spring-projects/spring-ai-examples/blob/main/model-context-protocol/weather/starter-stdio-server/src/main/java/org/springframework/ai/mcp/sample/server/WeatherService.java) that uses a REST client to query the data from the National Weather Service API:
873
874 ```java theme={null}
875 @Service
876 public class WeatherService {
877
878 private final RestClient restClient;
879
880 public WeatherService() {
881 this.restClient = RestClient.builder()
882 .baseUrl("https://api.weather.gov")
883 .defaultHeader("Accept", "application/geo+json")
884 .defaultHeader("User-Agent", "WeatherApiClient/1.0 ([email protected])")
885 .build();
886 }
887
888 @Tool(description = "Get weather forecast for a specific latitude/longitude")
889 public String getWeatherForecastByLocation(
890 double latitude, // Latitude coordinate
891 double longitude // Longitude coordinate
892 ) {
893 // Returns detailed forecast including:
894 // - Temperature and unit
895 // - Wind speed and direction
896 // - Detailed forecast description
897 }
898
899 @Tool(description = "Get weather alerts for a US state")
900 public String getAlerts(
901 @ToolParam(description = "Two-letter US state code (e.g. CA, NY)") String state
902 ) {
903 // Returns active alerts including:
904 // - Event type
905 // - Affected area
906 // - Severity
907 // - Description
908 // - Safety instructions
909 }
910
911 // ......
912 }
913 ```
914
915 The `@Service` annotation will auto-register the service in your application context.
916 The Spring AI `@Tool` annotation makes it easy to create and maintain MCP tools.
917
918 The auto-configuration will automatically register these tools with the MCP server.
919
920 ### Create your Boot Application
921
922 ```java theme={null}
923 @SpringBootApplication
924 public class McpServerApplication {
925
926 public static void main(String[] args) {
927 SpringApplication.run(McpServerApplication.class, args);
928 }
929
930 @Bean
931 public ToolCallbackProvider weatherTools(WeatherService weatherService) {
932 return MethodToolCallbackProvider.builder().toolObjects(weatherService).build();
933 }
934 }
935 ```
936
937 Uses the `MethodToolCallbackProvider` utils to convert the `@Tools` into actionable callbacks used by the MCP server.
938
939 ### Running the server
940
941 Finally, let's build the server:
942
943 ```bash theme={null}
944 ./mvnw clean install
945 ```
946
947 This will generate an `mcp-weather-stdio-server-0.0.1-SNAPSHOT.jar` file within the `target` folder.
948
949 Let's now test your server from an existing MCP host, Claude for Desktop.
950
951 ## Testing your server with Claude for Desktop
952
953 <Note>
954 Claude for Desktop is not yet available on Linux.
955 </Note>
956
957 First, make sure you have Claude for Desktop installed.
958 [You can install the latest version here.](https://claude.ai/download) If you already have Claude for Desktop, **make sure it's updated to the latest version.**
959
960 We'll need to configure Claude for Desktop for whichever MCP servers you want to use.
961 To do this, open your Claude for Desktop App configuration at `~/Library/Application Support/Claude/claude_desktop_config.json` in a text editor.
962 Make sure to create the file if it doesn't exist.
963
964 For example, if you have [VS Code](https://code.visualstudio.com/) installed:
965
966 <CodeGroup>
967 ```bash macOS/Linux theme={null}
968 code ~/Library/Application\ Support/Claude/claude_desktop_config.json
969 ```
970
971 ```powershell Windows theme={null}
972 code $env:AppData\Claude\claude_desktop_config.json
973 ```
974 </CodeGroup>
975
976 You'll then add your servers in the `mcpServers` key.
977 The MCP UI elements will only show up in Claude for Desktop if at least one server is properly configured.
978
979 In this case, we'll add our single weather server like so:
980
981 <CodeGroup>
982 ```json macOS/Linux theme={null}
983 {
984 "mcpServers": {
985 "spring-ai-mcp-weather": {
986 "command": "java",
987 "args": [
988 "-Dspring.ai.mcp.server.stdio=true",
989 "-jar",
990 "/ABSOLUTE/PATH/TO/PARENT/FOLDER/mcp-weather-stdio-server-0.0.1-SNAPSHOT.jar"
991 ]
992 }
993 }
994 }
995 ```
996
997 ```json Windows theme={null}
998 {
999 "mcpServers": {
1000 "spring-ai-mcp-weather": {
1001 "command": "java",
1002 "args": [
1003 "-Dspring.ai.mcp.server.transport=STDIO",
1004 "-jar",
1005 "C:\\ABSOLUTE\\PATH\\TO\\PARENT\\FOLDER\\weather\\mcp-weather-stdio-server-0.0.1-SNAPSHOT.jar"
1006 ]
1007 }
1008 }
1009 }
1010 ```
1011 </CodeGroup>
1012
1013 <Note>
1014 Make sure you pass in the absolute path to your server.
1015 </Note>
1016
1017 This tells Claude for Desktop:
1018
1019 1. There's an MCP server named "my-weather-server"
1020 2. To launch it by running `java -jar /ABSOLUTE/PATH/TO/PARENT/FOLDER/mcp-weather-stdio-server-0.0.1-SNAPSHOT.jar`
1021
1022 Save the file, and restart **Claude for Desktop**.
1023
1024 ## Testing your server with Java client
1025
1026 ### Create an MCP Client manually
1027
1028 Use the `McpClient` to connect to the server:
1029
1030 ```java theme={null}
1031 var stdioParams = ServerParameters.builder("java")
1032 .args("-jar", "/ABSOLUTE/PATH/TO/PARENT/FOLDER/mcp-weather-stdio-server-0.0.1-SNAPSHOT.jar")
1033 .build();
1034
1035 var stdioTransport = new StdioClientTransport(stdioParams);
1036
1037 var mcpClient = McpClient.sync(stdioTransport).build();
1038
1039 mcpClient.initialize();
1040
1041 ListToolsResult toolsList = mcpClient.listTools();
1042
1043 CallToolResult weather = mcpClient.callTool(
1044 new CallToolRequest("getWeatherForecastByLocation",
1045 Map.of("latitude", "47.6062", "longitude", "-122.3321")));
1046
1047 CallToolResult alert = mcpClient.callTool(
1048 new CallToolRequest("getAlerts", Map.of("state", "NY")));
1049
1050 mcpClient.closeGracefully();
1051 ```
1052
1053 ### Use MCP Client Boot Starter
1054
1055 Create a new boot starter application using the `spring-ai-starter-mcp-client` dependency:
1056
1057 ```xml theme={null}
1058 <dependency>
1059 <groupId>org.springframework.ai</groupId>
1060 <artifactId>spring-ai-starter-mcp-client</artifactId>
1061 </dependency>
1062 ```
1063
1064 and set the `spring.ai.mcp.client.stdio.servers-configuration` property to point to your `claude_desktop_config.json`.
1065 You can reuse the existing Anthropic Desktop configuration:
1066
1067 ```properties theme={null}
1068 spring.ai.mcp.client.stdio.servers-configuration=file:PATH/TO/claude_desktop_config.json
1069 ```
1070
1071 When you start your client application, the auto-configuration will automatically create MCP clients from the claude\_desktop\_config.json.
1072
1073 For more information, see the [MCP Client Boot Starters](https://docs.spring.io/spring-ai/reference/api/mcp/mcp-client-boot-starter-docs.html) reference documentation.
1074
1075 ## More Java MCP Server examples
1076
1077 The [starter-webflux-server](https://github.com/spring-projects/spring-ai-examples/tree/main/model-context-protocol/weather/starter-webflux-server) demonstrates how to create an MCP server using SSE transport.
1078 It showcases how to define and register MCP Tools, Resources, and Prompts, using the Spring Boot's auto-configuration capabilities.
1079 </Tab>
1080
1081 <Tab title="Kotlin">
1082 Let's get started with building our weather server! [You can find the complete code for what we'll be building here.](https://github.com/modelcontextprotocol/kotlin-sdk/tree/main/samples/weather-stdio-server)
1083
1084 ### Prerequisite knowledge
1085
1086 This quickstart assumes you have familiarity with:
1087
1088 * Kotlin
1089 * LLMs like Claude
1090
1091 ### Logging in MCP Servers
1092
1093 When implementing MCP servers, be careful about how you handle logging:
1094
1095 **For STDIO-based servers:** Never use `println()`, as it writes to standard output (stdout) by default. Writing to stdout will corrupt the JSON-RPC messages and break your server.
1096
1097 **For HTTP-based servers:** Standard output logging is fine since it doesn't interfere with HTTP responses.
1098
1099 ### Best Practices
1100
1101 * Use a logging library that writes to stderr or files.
1102
1103 ### System requirements
1104
1105 * JDK 11 or higher installed.
1106
1107 ### Set up your environment
1108
1109 First, let's install `java` and `gradle` if you haven't already.
1110 You can download `java` from [official Oracle JDK website](https://www.oracle.com/java/technologies/downloads/).
1111 Verify your `java` installation:
1112
1113 ```bash theme={null}
1114 java --version
1115 ```
1116
1117 Now, let's create and set up your project:
1118
1119 <CodeGroup>
1120 ```bash macOS/Linux theme={null}
1121 # Create a new directory for our project
1122 mkdir weather
1123 cd weather
1124
1125 # Initialize a new kotlin project
1126 gradle init
1127 ```
1128
1129 ```powershell Windows theme={null}
1130 # Create a new directory for our project
1131 md weather
1132 cd weather
1133
1134 # Initialize a new kotlin project
1135 gradle init
1136 ```
1137 </CodeGroup>
1138
1139 After running `gradle init`, select **Application** as the project type, **Kotlin** as the programming language.
1140
1141 Alternatively, you can create a Kotlin application using the [IntelliJ IDEA project wizard](https://kotlinlang.org/docs/jvm-get-started.html).
1142
1143 After creating the project, replace the contents of your `build.gradle.kts` with:
1144
1145 ```kotlin build.gradle.kts theme={null}
1146 // Check latest versions at https://github.com/modelcontextprotocol/kotlin-sdk/releases
1147 val mcpVersion = "0.9.0"
1148 val ktorVersion = "3.2.3"
1149 val slf4jVersion = "2.0.17"
1150
1151 plugins {
1152 kotlin("jvm") version "2.3.20"
1153 kotlin("plugin.serialization") version "2.3.20"
1154 id("com.gradleup.shadow") version "8.3.9"
1155 application
1156 }
1157
1158 application {
1159 mainClass.set("MainKt")
1160 }
1161
1162 dependencies {
1163 implementation("io.modelcontextprotocol:kotlin-sdk:$mcpVersion")
1164 implementation("io.ktor:ktor-client-content-negotiation:$ktorVersion")
1165 implementation("io.ktor:ktor-serialization-kotlinx-json:$ktorVersion")
1166 implementation("io.ktor:ktor-client-cio:$ktorVersion")
1167 implementation("org.slf4j:slf4j-simple:$slf4jVersion")
1168 }
1169 ```
1170
1171 Verify that everything is set up correctly:
1172
1173 ```bash theme={null}
1174 ./gradlew build
1175 ```
1176
1177 Now let's dive into building your server.
1178
1179 ## Building your server
1180
1181 ### Setting up the instance
1182
1183 Add a server initialization function:
1184
1185 ```kotlin theme={null}
1186 fun runMcpServer() {
1187 val server = Server(
1188 Implementation(
1189 name = "weather",
1190 version = "1.0.0",
1191 ),
1192 ServerOptions(
1193 capabilities = ServerCapabilities(tools = ServerCapabilities.Tools(listChanged = true)),
1194 ),
1195 )
1196
1197 // register tools on server here
1198
1199 val transport = StdioServerTransport(
1200 System.`in`.asInput(),
1201 System.out.asSink().buffered(),
1202 )
1203
1204 runBlocking {
1205 val session = server.createSession(transport)
1206 val done = Job()
1207 session.onClose {
1208 done.complete()
1209 }
1210 done.join()
1211 }
1212 }
1213 ```
1214
1215 ### Weather API helper functions
1216
1217 Next, let's add functions and data classes for querying and converting responses from the National Weather Service API:
1218
1219 ```kotlin theme={null}
1220 val httpClient = HttpClient(CIO) {
1221 defaultRequest {
1222 url("https://api.weather.gov")
1223 headers {
1224 append("Accept", "application/geo+json")
1225 append("User-Agent", "WeatherApiClient/1.0")
1226 }
1227 contentType(ContentType.Application.Json)
1228 }
1229 install(ContentNegotiation) {
1230 json(Json { ignoreUnknownKeys = true })
1231 }
1232 }
1233
1234 // Extension function to fetch weather alerts for a given state
1235 suspend fun HttpClient.getAlerts(state: String): List<String> {
1236 val alerts = this.get("/alerts/active/area/$state").body<AlertsResponse>()
1237 return alerts.features.map { feature ->
1238 """
1239 Event: ${feature.properties.event}
1240 Area: ${feature.properties.areaDesc}
1241 Severity: ${feature.properties.severity}
1242 Status: ${feature.properties.status}
1243 Headline: ${feature.properties.headline}
1244 """.trimIndent()
1245 }
1246 }
1247
1248 // Extension function to fetch forecast information for given latitude and longitude
1249 suspend fun HttpClient.getForecast(latitude: Double, longitude: Double): List<String> {
1250 val points = this.get("/points/$latitude,$longitude").body<PointsResponse>()
1251 val forecastUrl = points.properties.forecast ?: error("No forecast URL available")
1252 val forecast = this.get(forecastUrl).body<ForecastResponse>()
1253 return forecast.properties.periods.map { period ->
1254 """
1255 ${period.name}:
1256 Temperature: ${period.temperature}°${period.temperatureUnit}
1257 Wind: ${period.windSpeed} ${period.windDirection}
1258 ${period.shortForecast}
1259 """.trimIndent()
1260 }
1261 }
1262
1263 @Serializable
1264 data class PointsResponse(val properties: PointsProperties)
1265
1266 @Serializable
1267 data class PointsProperties(val forecast: String? = null)
1268
1269 @Serializable
1270 data class ForecastResponse(val properties: ForecastProperties)
1271
1272 @Serializable
1273 data class ForecastProperties(val periods: List<ForecastPeriod> = emptyList())
1274
1275 @Serializable
1276 data class ForecastPeriod(
1277 val name: String? = null,
1278 val temperature: Int? = null,
1279 val temperatureUnit: String? = null,
1280 val windSpeed: String? = null,
1281 val windDirection: String? = null,
1282 val shortForecast: String? = null,
1283 )
1284
1285 @Serializable
1286 data class AlertsResponse(val features: List<AlertFeature> = emptyList())
1287
1288 @Serializable
1289 data class AlertFeature(val properties: AlertProperties)
1290
1291 @Serializable
1292 data class AlertProperties(
1293 val event: String? = null,
1294 val areaDesc: String? = null,
1295 val severity: String? = null,
1296 val status: String? = null,
1297 val headline: String? = null,
1298 )
1299 ```
1300
1301 ### Implementing tool execution
1302
1303 The tool execution handler is responsible for actually executing the logic of each tool. Let's add it:
1304
1305 ```kotlin theme={null}
1306 // Register weather tools
1307
1308 server.addTool(
1309 name = "get_alerts",
1310 description = "Get weather alerts for a US state. Input is a two-letter US state code (e.g. CA, NY)",
1311 inputSchema = ToolSchema(
1312 properties = buildJsonObject {
1313 putJsonObject("state") {
1314 put("type", "string")
1315 put("description", "Two-letter US state code (e.g. CA, NY)")
1316 }
1317 },
1318 required = listOf("state"),
1319 ),
1320 ) { request ->
1321 val state = request.arguments?.get("state")?.jsonPrimitive?.content
1322 ?: return@addTool CallToolResult(
1323 content = listOf(TextContent("The 'state' parameter is required.")),
1324 )
1325
1326 val alerts = httpClient.getAlerts(state)
1327 CallToolResult(content = alerts.map { TextContent(it) })
1328 }
1329
1330 server.addTool(
1331 name = "get_forecast",
1332 description = "Get weather forecast for a location. Note: only US locations are supported by the NWS API.",
1333 inputSchema = ToolSchema(
1334 properties = buildJsonObject {
1335 putJsonObject("latitude") {
1336 put("type", "number")
1337 put("description", "Latitude of the location")
1338 }
1339 putJsonObject("longitude") {
1340 put("type", "number")
1341 put("description", "Longitude of the location")
1342 }
1343 },
1344 required = listOf("latitude", "longitude"),
1345 ),
1346 ) { request ->
1347 val latitude = request.arguments?.get("latitude")?.jsonPrimitive?.doubleOrNull
1348 val longitude = request.arguments?.get("longitude")?.jsonPrimitive?.doubleOrNull
1349 if (latitude == null || longitude == null) {
1350 return@addTool CallToolResult(
1351 content = listOf(TextContent("The 'latitude' and 'longitude' parameters are required.")),
1352 )
1353 }
1354
1355 val forecast = httpClient.getForecast(latitude, longitude)
1356 CallToolResult(content = forecast.map { TextContent(it) })
1357 }
1358 ```
1359
1360 ### Running the server
1361
1362 Finally, implement the main function to run the server:
1363
1364 ```kotlin theme={null}
1365 fun main() = runMcpServer()
1366 ```
1367
1368 You can run the server directly during development:
1369
1370 ```bash theme={null}
1371 ./gradlew run
1372 ```
1373
1374 For production use, build the shadow JAR:
1375
1376 ```bash theme={null}
1377 ./gradlew build
1378 java -jar build/libs/weather-0.1.0-all.jar
1379 ```
1380
1381 Let's now test your server from an existing MCP host, Claude for Desktop.
1382
1383 ## Testing your server with Claude for Desktop
1384
1385 <Note>
1386 Claude for Desktop is not yet available on Linux. Linux users can proceed to the [Building a client](/docs/2024-11-05/develop/build-client) tutorial to build an MCP client that connects to the server we just built.
1387 </Note>
1388
1389 First, make sure you have Claude for Desktop installed. [You can install the latest version
1390 here.](https://claude.ai/download) If you already have Claude for Desktop, **make sure it's updated to the latest version.**
1391
1392 We'll need to configure Claude for Desktop for whichever MCP servers you want to use.
1393 To do this, open your Claude for Desktop App configuration at `~/Library/Application Support/Claude/claude_desktop_config.json` in a text editor.
1394 Make sure to create the file if it doesn't exist.
1395
1396 For example, if you have [VS Code](https://code.visualstudio.com/) installed:
1397
1398 <CodeGroup>
1399 ```bash macOS/Linux theme={null}
1400 code ~/Library/Application\ Support/Claude/claude_desktop_config.json
1401 ```
1402
1403 ```powershell Windows theme={null}
1404 code $env:AppData\Claude\claude_desktop_config.json
1405 ```
1406 </CodeGroup>
1407
1408 You'll then add your servers in the `mcpServers` key.
1409 The MCP UI elements will only show up in Claude for Desktop if at least one server is properly configured.
1410
1411 In this case, we'll add our single weather server like so:
1412
1413 <CodeGroup>
1414 ```json macOS/Linux theme={null}
1415 {
1416 "mcpServers": {
1417 "weather": {
1418 "command": "java",
1419 "args": [
1420 "-jar",
1421 "/ABSOLUTE/PATH/TO/PARENT/FOLDER/weather/build/libs/weather-0.1.0-all.jar"
1422 ]
1423 }
1424 }
1425 }
1426 ```
1427
1428 ```json Windows theme={null}
1429 {
1430 "mcpServers": {
1431 "weather": {
1432 "command": "java",
1433 "args": [
1434 "-jar",
1435 "C:\\PATH\\TO\\PARENT\\FOLDER\\weather\\build\\libs\\weather-0.1.0-all.jar"
1436 ]
1437 }
1438 }
1439 }
1440 ```
1441 </CodeGroup>
1442
1443 This tells Claude for Desktop:
1444
1445 1. There's an MCP server named "weather"
1446 2. Launch it by running `java -jar /ABSOLUTE/PATH/TO/PARENT/FOLDER/weather/build/libs/weather-0.1.0-all.jar`
1447
1448 Save the file, and restart **Claude for Desktop**.
1449 </Tab>
1450
1451 <Tab title="C#">
1452 Let's get started with building our weather server! [You can find the complete code for what we'll be building here.](https://github.com/modelcontextprotocol/csharp-sdk/tree/main/samples/QuickstartWeatherServer)
1453
1454 ### Prerequisite knowledge
1455
1456 This quickstart assumes you have familiarity with:
1457
1458 * C#
1459 * LLMs like Claude
1460 * .NET 8 or higher
1461
1462 ### Logging in MCP Servers
1463
1464 When implementing MCP servers, be careful about how you handle logging:
1465
1466 **For STDIO-based servers:** Never use `Console.WriteLine()` or `Console.Write()`, as they write to standard output (stdout). Writing to stdout will corrupt the JSON-RPC messages and break your server.
1467
1468 **For HTTP-based servers:** Standard output logging is fine since it doesn't interfere with HTTP responses.
1469
1470 ### Best Practices
1471
1472 * Use a logging library that writes to stderr or files.
1473
1474 ### System requirements
1475
1476 * [.NET 8 SDK](https://dotnet.microsoft.com/download/dotnet/8.0) or higher installed.
1477
1478 ### Set up your environment
1479
1480 First, let's install `dotnet` if you haven't already. You can download `dotnet` from [official Microsoft .NET website](https://dotnet.microsoft.com/download/). Verify your `dotnet` installation:
1481
1482 ```bash theme={null}
1483 dotnet --version
1484 ```
1485
1486 Now, let's create and set up your project:
1487
1488 <CodeGroup>
1489 ```bash macOS/Linux theme={null}
1490 # Create a new directory for our project
1491 mkdir weather
1492 cd weather
1493 # Initialize a new C# project
1494 dotnet new console
1495 ```
1496
1497 ```powershell Windows theme={null}
1498 # Create a new directory for our project
1499 mkdir weather
1500 cd weather
1501 # Initialize a new C# project
1502 dotnet new console
1503 ```
1504 </CodeGroup>
1505
1506 After running `dotnet new console`, you will be presented with a new C# project.
1507 You can open the project in your favorite IDE, such as [Visual Studio](https://visualstudio.microsoft.com/) or [Rider](https://www.jetbrains.com/rider/).
1508 Alternatively, you can create a C# application using the [Visual Studio project wizard](https://learn.microsoft.com/en-us/visualstudio/get-started/csharp/tutorial-console?view=vs-2022).
1509 After creating the project, add NuGet package for the Model Context Protocol SDK and hosting:
1510
1511 ```bash theme={null}
1512 # Add the Model Context Protocol SDK NuGet package
1513 dotnet add package ModelContextProtocol --prerelease
1514 # Add the .NET Hosting NuGet package
1515 dotnet add package Microsoft.Extensions.Hosting
1516 ```
1517
1518 Now let’s dive into building your server.
1519
1520 ## Building your server
1521
1522 Open the `Program.cs` file in your project and replace its contents with the following code:
1523
1524 ```csharp theme={null}
1525 using Microsoft.Extensions.DependencyInjection;
1526 using Microsoft.Extensions.Hosting;
1527 using ModelContextProtocol;
1528 using System.Net.Http.Headers;
1529
1530 var builder = Host.CreateEmptyApplicationBuilder(settings: null);
1531
1532 builder.Services.AddMcpServer()
1533 .WithStdioServerTransport()
1534 .WithToolsFromAssembly();
1535
1536 builder.Services.AddSingleton(_ =>
1537 {
1538 var client = new HttpClient() { BaseAddress = new Uri("https://api.weather.gov") };
1539 client.DefaultRequestHeaders.UserAgent.Add(new ProductInfoHeaderValue("weather-tool", "1.0"));
1540 return client;
1541 });
1542
1543 var app = builder.Build();
1544
1545 await app.RunAsync();
1546 ```
1547
1548 <Note>
1549 When creating the `ApplicationHostBuilder`, ensure you use `CreateEmptyApplicationBuilder` instead of `CreateDefaultBuilder`. This ensures that the server does not write any additional messages to the console. This is only necessary for servers using STDIO transport.
1550 </Note>
1551
1552 This code sets up a basic console application that uses the Model Context Protocol SDK to create an MCP server with standard I/O transport.
1553
1554 ### Weather API helper functions
1555
1556 Create an extension class for `HttpClient` which helps simplify JSON request handling:
1557
1558 ```csharp theme={null}
1559 using System.Text.Json;
1560
1561 internal static class HttpClientExt
1562 {
1563 public static async Task<JsonDocument> ReadJsonDocumentAsync(this HttpClient client, string requestUri)
1564 {
1565 using var response = await client.GetAsync(requestUri);
1566 response.EnsureSuccessStatusCode();
1567 return await JsonDocument.ParseAsync(await response.Content.ReadAsStreamAsync());
1568 }
1569 }
1570 ```
1571
1572 Next, define a class with the tool execution handlers for querying and converting responses from the National Weather Service API:
1573
1574 ```csharp theme={null}
1575 using ModelContextProtocol.Server;
1576 using System.ComponentModel;
1577 using System.Globalization;
1578 using System.Text.Json;
1579
1580 namespace QuickstartWeatherServer.Tools;
1581
1582 [McpServerToolType]
1583 public static class WeatherTools
1584 {
1585 [McpServerTool, Description("Get weather alerts for a US state code.")]
1586 public static async Task<string> GetAlerts(
1587 HttpClient client,
1588 [Description("The US state code to get alerts for.")] string state)
1589 {
1590 using var jsonDocument = await client.ReadJsonDocumentAsync($"/alerts/active/area/{state}");
1591 var jsonElement = jsonDocument.RootElement;
1592 var alerts = jsonElement.GetProperty("features").EnumerateArray();
1593
1594 if (!alerts.Any())
1595 {
1596 return "No active alerts for this state.";
1597 }
1598
1599 return string.Join("\n--\n", alerts.Select(alert =>
1600 {
1601 JsonElement properties = alert.GetProperty("properties");
1602 return $"""
1603 Event: {properties.GetProperty("event").GetString()}
1604 Area: {properties.GetProperty("areaDesc").GetString()}
1605 Severity: {properties.GetProperty("severity").GetString()}
1606 Description: {properties.GetProperty("description").GetString()}
1607 Instruction: {properties.GetProperty("instruction").GetString()}
1608 """;
1609 }));
1610 }
1611
1612 [McpServerTool, Description("Get weather forecast for a location.")]
1613 public static async Task<string> GetForecast(
1614 HttpClient client,
1615 [Description("Latitude of the location.")] double latitude,
1616 [Description("Longitude of the location.")] double longitude)
1617 {
1618 var pointUrl = string.Create(CultureInfo.InvariantCulture, $"/points/{latitude},{longitude}");
1619 using var jsonDocument = await client.ReadJsonDocumentAsync(pointUrl);
1620 var forecastUrl = jsonDocument.RootElement.GetProperty("properties").GetProperty("forecast").GetString()
1621 ?? throw new Exception($"No forecast URL provided by {client.BaseAddress}points/{latitude},{longitude}");
1622
1623 using var forecastDocument = await client.ReadJsonDocumentAsync(forecastUrl);
1624 var periods = forecastDocument.RootElement.GetProperty("properties").GetProperty("periods").EnumerateArray();
1625
1626 return string.Join("\n---\n", periods.Select(period => $"""
1627 {period.GetProperty("name").GetString()}
1628 Temperature: {period.GetProperty("temperature").GetInt32()}°F
1629 Wind: {period.GetProperty("windSpeed").GetString()} {period.GetProperty("windDirection").GetString()}
1630 Forecast: {period.GetProperty("detailedForecast").GetString()}
1631 """));
1632 }
1633 }
1634 ```
1635
1636 ### Running the server
1637
1638 Finally, run the server using the following command:
1639
1640 ```bash theme={null}
1641 dotnet run
1642 ```
1643
1644 This will start the server and listen for incoming requests on standard input/output.
1645
1646 ## Testing your server with Claude for Desktop
1647
1648 <Note>
1649 Claude for Desktop is not yet available on Linux. Linux users can proceed to the [Building a client](/docs/2024-11-05/develop/build-client) tutorial to build an MCP client that connects to the server we just built.
1650 </Note>
1651
1652 First, make sure you have Claude for Desktop installed. [You can install the latest version
1653 here.](https://claude.ai/download) If you already have Claude for Desktop, **make sure it's updated to the latest version.**
1654 We'll need to configure Claude for Desktop for whichever MCP servers you want to use. To do this, open your Claude for Desktop App configuration at `~/Library/Application Support/Claude/claude_desktop_config.json` in a text editor. Make sure to create the file if it doesn't exist.
1655 For example, if you have [VS Code](https://code.visualstudio.com/) installed:
1656
1657 <CodeGroup>
1658 ```bash macOS/Linux theme={null}
1659 code ~/Library/Application\ Support/Claude/claude_desktop_config.json
1660 ```
1661
1662 ```powershell Windows theme={null}
1663 code $env:AppData\Claude\claude_desktop_config.json
1664 ```
1665 </CodeGroup>
1666
1667 You'll then add your servers in the `mcpServers` key. The MCP UI elements will only show up in Claude for Desktop if at least one server is properly configured.
1668 In this case, we'll add our single weather server like so:
1669
1670 <CodeGroup>
1671 ```json macOS/Linux theme={null}
1672 {
1673 "mcpServers": {
1674 "weather": {
1675 "command": "dotnet",
1676 "args": ["run", "--project", "/ABSOLUTE/PATH/TO/PROJECT", "--no-build"]
1677 }
1678 }
1679 }
1680 ```
1681
1682 ```json Windows theme={null}
1683 {
1684 "mcpServers": {
1685 "weather": {
1686 "command": "dotnet",
1687 "args": [
1688 "run",
1689 "--project",
1690 "C:\\ABSOLUTE\\PATH\\TO\\PROJECT",
1691 "--no-build"
1692 ]
1693 }
1694 }
1695 }
1696 ```
1697 </CodeGroup>
1698
1699 This tells Claude for Desktop:
1700
1701 1. There's an MCP server named "weather"
1702 2. Launch it by running `dotnet run /ABSOLUTE/PATH/TO/PROJECT`
1703 Save the file, and restart **Claude for Desktop**.
1704 </Tab>
1705
1706 <Tab title="Ruby">
1707 Let's get started with building our weather server! [You can find the complete code for what we'll be building here.](https://github.com/modelcontextprotocol/quickstart-resources/tree/main/weather-server-ruby)
1708
1709 ### Prerequisite knowledge
1710
1711 This quickstart assumes you have familiarity with:
1712
1713 * Ruby
1714 * LLMs like Claude
1715
1716 ### Logging in MCP Servers
1717
1718 When implementing MCP servers, be careful about how you handle logging:
1719
1720 **For STDIO-based servers:** Never use `puts` or `print`, as they write to standard output (stdout) by default. Writing to stdout will corrupt the JSON-RPC messages and break your server.
1721
1722 **For HTTP-based servers:** Standard output logging is fine since it doesn't interfere with HTTP responses.
1723
1724 ### Best Practices
1725
1726 * Use a logging library that writes to stderr or files.
1727
1728 ### Quick Examples
1729
1730 ```ruby theme={null}
1731 # ❌ Bad (STDIO)
1732 puts "Processing request"
1733
1734 # ✅ Good (STDIO)
1735 require "logger"
1736 logger = Logger.new($stderr)
1737 logger.info("Processing request")
1738 ```
1739
1740 ### System requirements
1741
1742 * Ruby 2.7 or higher installed.
1743
1744 ### Set up your environment
1745
1746 First, let's make sure you have Ruby installed. You can check by running:
1747
1748 ```bash theme={null}
1749 ruby --version
1750 ```
1751
1752 Now, let's create and set up our project:
1753
1754 <CodeGroup>
1755 ```bash macOS/Linux theme={null}
1756 # Create a new directory for our project
1757 mkdir weather
1758 cd weather
1759
1760 # Create a Gemfile
1761 bundle init
1762
1763 # Add the MCP SDK dependency
1764 bundle add mcp
1765
1766 # Create our server file
1767 touch weather.rb
1768 ```
1769
1770 ```powershell Windows theme={null}
1771 # Create a new directory for our project
1772 mkdir weather
1773 cd weather
1774
1775 # Create a Gemfile
1776 bundle init
1777
1778 # Add the MCP SDK dependency
1779 bundle add mcp
1780
1781 # Create our server file
1782 new-item weather.rb
1783 ```
1784 </CodeGroup>
1785
1786 Now let's dive into building your server.
1787
1788 ## Building your server
1789
1790 ### Importing packages and setting up constants
1791
1792 Open `weather.rb` and add these requires and constants at the top:
1793
1794 ```ruby theme={null}
1795 require "json"
1796 require "mcp"
1797 require "net/http"
1798 require "uri"
1799
1800 NWS_API_BASE = "https://api.weather.gov"
1801 USER_AGENT = "weather-app/1.0"
1802 ```
1803
1804 The `mcp` gem provides the Model Context Protocol SDK for Ruby, with classes for server implementation and stdio transport.
1805
1806 ### Helper methods
1807
1808 Next, let's add helper methods for querying and formatting data from the National Weather Service API:
1809
1810 ```ruby theme={null}
1811 module HelperMethods
1812 def make_nws_request(url)
1813 uri = URI(url)
1814 request = Net::HTTP::Get.new(uri)
1815 request["User-Agent"] = USER_AGENT
1816 request["Accept"] = "application/geo+json"
1817
1818 response = Net::HTTP.start(uri.hostname, uri.port, use_ssl: true) do |http|
1819 http.request(request)
1820 end
1821
1822 raise "HTTP #{response.code}: #{response.message}" unless response.is_a?(Net::HTTPSuccess)
1823
1824 JSON.parse(response.body)
1825 end
1826
1827 def format_alert(feature)
1828 properties = feature["properties"]
1829
1830 <<~ALERT
1831 Event: #{properties["event"] || "Unknown"}
1832 Area: #{properties["areaDesc"] || "Unknown"}
1833 Severity: #{properties["severity"] || "Unknown"}
1834 Description: #{properties["description"] || "No description available"}
1835 Instructions: #{properties["instruction"] || "No specific instructions provided"}
1836 ALERT
1837 end
1838 end
1839 ```
1840
1841 ### Implementing tool execution
1842
1843 Now let's define our tool classes. Each tool subclasses `MCP::Tool` and implements the tool logic:
1844
1845 ```ruby theme={null}
1846 class GetAlerts < MCP::Tool
1847 extend HelperMethods
1848
1849 tool_name "get_alerts"
1850 description "Get weather alerts for a US state"
1851 input_schema(
1852 properties: {
1853 state: {
1854 type: "string",
1855 description: "Two-letter US state code (e.g. CA, NY)"
1856 }
1857 },
1858 required: ["state"]
1859 )
1860
1861 def self.call(state:)
1862 url = "#{NWS_API_BASE}/alerts/active/area/#{state.upcase}"
1863 data = make_nws_request(url)
1864
1865 if data["features"].empty?
1866 return MCP::Tool::Response.new([{
1867 type: "text",
1868 text: "No active alerts for this state."
1869 }])
1870 end
1871
1872 alerts = data["features"].map { |feature| format_alert(feature) }
1873 MCP::Tool::Response.new([{
1874 type: "text",
1875 text: alerts.join("\n---\n")
1876 }])
1877 end
1878 end
1879
1880 class GetForecast < MCP::Tool
1881 extend HelperMethods
1882
1883 tool_name "get_forecast"
1884 description "Get weather forecast for a location"
1885 input_schema(
1886 properties: {
1887 latitude: {
1888 type: "number",
1889 description: "Latitude of the location"
1890 },
1891 longitude: {
1892 type: "number",
1893 description: "Longitude of the location"
1894 }
1895 },
1896 required: ["latitude", "longitude"]
1897 )
1898
1899 def self.call(latitude:, longitude:)
1900 # First get the forecast grid endpoint.
1901 points_url = "#{NWS_API_BASE}/points/#{latitude},#{longitude}"
1902 points_data = make_nws_request(points_url)
1903
1904 # Get the forecast URL from the points response.
1905 forecast_url = points_data["properties"]["forecast"]
1906 forecast_data = make_nws_request(forecast_url)
1907
1908 # Format the periods into a readable forecast.
1909 periods = forecast_data["properties"]["periods"]
1910 forecasts = periods.first(5).map do |period|
1911 <<~FORECAST
1912 #{period["name"]}:
1913 Temperature: #{period["temperature"]}°#{period["temperatureUnit"]}
1914 Wind: #{period["windSpeed"]} #{period["windDirection"]}
1915 Forecast: #{period["detailedForecast"]}
1916 FORECAST
1917 end
1918
1919 MCP::Tool::Response.new([{
1920 type: "text",
1921 text: forecasts.join("\n---\n")
1922 }])
1923 end
1924 end
1925 ```
1926
1927 ### Running the server
1928
1929 Finally, initialize and run the server:
1930
1931 ```ruby theme={null}
1932 server = MCP::Server.new(
1933 name: "weather",
1934 version: "1.0.0",
1935 tools: [GetAlerts, GetForecast]
1936 )
1937
1938 transport = MCP::Server::Transports::StdioTransport.new(server)
1939 transport.open
1940 ```
1941
1942 Your server is complete! Run `bundle exec ruby weather.rb` to start the MCP server, which will listen for messages from MCP hosts.
1943
1944 Let's now test your server from an existing MCP host, Claude for Desktop.
1945
1946 ## Testing your server with Claude for Desktop
1947
1948 <Note>
1949 Claude for Desktop is not yet available on Linux. Linux users can proceed to the [Building a client](/docs/2024-11-05/develop/build-client) tutorial to build an MCP client that connects to the server we just built.
1950 </Note>
1951
1952 First, make sure you have Claude for Desktop installed. [You can install the latest version here.](https://claude.ai/download) If you already have Claude for Desktop, **make sure it's updated to the latest version.**
1953
1954 We'll need to configure Claude for Desktop for whichever MCP servers you want to use. To do this, open your Claude for Desktop App configuration at `~/Library/Application Support/Claude/claude_desktop_config.json` in a text editor. Make sure to create the file if it doesn't exist.
1955
1956 For example, if you have [VS Code](https://code.visualstudio.com/) installed:
1957
1958 <CodeGroup>
1959 ```bash macOS/Linux theme={null}
1960 code ~/Library/Application\ Support/Claude/claude_desktop_config.json
1961 ```
1962
1963 ```powershell Windows theme={null}
1964 code $env:AppData\Claude\claude_desktop_config.json
1965 ```
1966 </CodeGroup>
1967
1968 You'll then add your servers in the `mcpServers` key. The MCP UI elements will only show up in Claude for Desktop if at least one server is properly configured.
1969
1970 In this case, we'll add our single weather server like so:
1971
1972 <CodeGroup>
1973 ```json macOS/Linux theme={null}
1974 {
1975 "mcpServers": {
1976 "weather": {
1977 "command": "bundle",
1978 "args": ["exec", "ruby", "weather.rb"],
1979 "cwd": "/ABSOLUTE/PATH/TO/PARENT/FOLDER/weather"
1980 }
1981 }
1982 }
1983 ```
1984
1985 ```json Windows theme={null}
1986 {
1987 "mcpServers": {
1988 "weather": {
1989 "command": "bundle",
1990 "args": ["exec", "ruby", "weather.rb"],
1991 "cwd": "C:\\ABSOLUTE\\PATH\\TO\\PARENT\\FOLDER\\weather"
1992 }
1993 }
1994 }
1995 ```
1996 </CodeGroup>
1997
1998 <Note>
1999 Make sure you pass in the absolute path to your project directory in the `cwd` field. You can get this by running `pwd` on macOS/Linux or `cd` on Windows Command Prompt from your project directory. On Windows, remember to use double backslashes (`\\`) or forward slashes (`/`) in the JSON path.
2000 </Note>
2001
2002 This tells Claude for Desktop:
2003
2004 1. There's an MCP server named "weather"
2005 2. Launch it by running `bundle exec ruby weather.rb` in the specified directory
2006
2007 Save the file, and restart **Claude for Desktop**.
2008 </Tab>
2009
2010 <Tab title="Rust">
2011 Let's get started with building our weather server! [You can find the complete code for what we'll be building here.](https://github.com/modelcontextprotocol/quickstart-resources/tree/main/weather-server-rust)
2012
2013 ### Prerequisite knowledge
2014
2015 This quickstart assumes you have familiarity with:
2016
2017 * Rust programming language
2018 * Async/await in Rust
2019 * LLMs like Claude
2020
2021 ### Logging in MCP Servers
2022
2023 When implementing MCP servers, be careful about how you handle logging:
2024
2025 **For STDIO-based servers:** Never use `println!()` or `print!()`, as they write to standard output (stdout). Writing to stdout will corrupt the JSON-RPC messages and break your server.
2026
2027 **For HTTP-based servers:** Standard output logging is fine since it doesn't interfere with HTTP responses.
2028
2029 ### Best Practices
2030
2031 * Use a logging library that writes to stderr or files, such as `tracing` or `log` in Rust.
2032 * Configure your logging framework to avoid stdout output.
2033
2034 ### Quick Examples
2035
2036 ```rust theme={null}
2037 // ❌ Bad (STDIO)
2038 println!("Processing request");
2039
2040 // ✅ Good (STDIO)
2041 eprintln!("Processing request"); // writes to stderr
2042 ```
2043
2044 ### System requirements
2045
2046 * Rust 1.70 or higher installed.
2047 * Cargo (comes with Rust installation).
2048
2049 ### Set up your environment
2050
2051 First, let's install Rust if you haven't already. You can install Rust from [rust-lang.org](https://www.rust-lang.org/tools/install):
2052
2053 <CodeGroup>
2054 ```bash macOS/Linux theme={null}
2055 curl --proto '=https' --tlsv1.2 -sSf https://sh.rustup.rs | sh
2056 ```
2057
2058 ```powershell Windows theme={null}
2059 # Download and run rustup-init.exe from https://rustup.rs/
2060 ```
2061 </CodeGroup>
2062
2063 Verify your Rust installation:
2064
2065 ```bash theme={null}
2066 rustc --version
2067 cargo --version
2068 ```
2069
2070 Now, let's create and set up our project:
2071
2072 <CodeGroup>
2073 ```bash macOS/Linux theme={null}
2074 # Create a new Rust project
2075 cargo new weather
2076 cd weather
2077 ```
2078
2079 ```powershell Windows theme={null}
2080 # Create a new Rust project
2081 cargo new weather
2082 cd weather
2083 ```
2084 </CodeGroup>
2085
2086 Update your `Cargo.toml` to add the required dependencies:
2087
2088 ```toml Cargo.toml theme={null}
2089 [package]
2090 name = "weather"
2091 version = "0.1.0"
2092 edition = "2024"
2093
2094 [dependencies]
2095 rmcp = { version = "0.3", features = ["server", "macros", "transport-io"] }
2096 tokio = { version = "1.46", features = ["full"] }
2097 reqwest = { version = "0.12", features = ["json"] }
2098 serde = { version = "1.0", features = ["derive"] }
2099 serde_json = "1.0"
2100 anyhow = "1.0"
2101 tracing = "0.1"
2102 tracing-subscriber = { version = "0.3", features = ["env-filter", "std", "fmt"] }
2103 ```
2104
2105 Now let's dive into building your server.
2106
2107 ## Building your server
2108
2109 ### Importing packages and constants
2110
2111 Open `src/main.rs` and add these imports and constants at the top:
2112
2113 ```rust theme={null}
2114 use anyhow::Result;
2115 use rmcp::{
2116 ServerHandler, ServiceExt,
2117 handler::server::{router::tool::ToolRouter, tool::Parameters},
2118 model::*,
2119 schemars, tool, tool_handler, tool_router,
2120 };
2121 use serde::Deserialize;
2122 use serde::de::DeserializeOwned;
2123
2124 const NWS_API_BASE: &str = "https://api.weather.gov";
2125 const USER_AGENT: &str = "weather-app/1.0";
2126 ```
2127
2128 The `rmcp` crate provides the Model Context Protocol SDK for Rust, with features for server implementation, procedural macros, and stdio transport.
2129
2130 ### Data structures
2131
2132 Next, let's define the data structures for deserializing responses from the National Weather Service API:
2133
2134 ```rust theme={null}
2135 #[derive(Debug, Deserialize)]
2136 struct AlertsResponse {
2137 features: Vec<AlertFeature>,
2138 }
2139
2140 #[derive(Debug, Deserialize)]
2141 struct AlertFeature {
2142 properties: AlertProperties,
2143 }
2144
2145 #[derive(Debug, Deserialize)]
2146 struct AlertProperties {
2147 event: Option<String>,
2148 #[serde(rename = "areaDesc")]
2149 area_desc: Option<String>,
2150 severity: Option<String>,
2151 description: Option<String>,
2152 instruction: Option<String>,
2153 }
2154
2155 #[derive(Debug, Deserialize)]
2156 struct PointsResponse {
2157 properties: PointsProperties,
2158 }
2159
2160 #[derive(Debug, Deserialize)]
2161 struct PointsProperties {
2162 forecast: String,
2163 }
2164
2165 #[derive(Debug, Deserialize)]
2166 struct ForecastResponse {
2167 properties: ForecastProperties,
2168 }
2169
2170 #[derive(Debug, Deserialize)]
2171 struct ForecastProperties {
2172 periods: Vec<ForecastPeriod>,
2173 }
2174
2175 #[derive(Debug, Deserialize)]
2176 struct ForecastPeriod {
2177 name: String,
2178 temperature: i32,
2179 #[serde(rename = "temperatureUnit")]
2180 temperature_unit: String,
2181 #[serde(rename = "windSpeed")]
2182 wind_speed: String,
2183 #[serde(rename = "windDirection")]
2184 wind_direction: String,
2185 #[serde(rename = "detailedForecast")]
2186 detailed_forecast: String,
2187 }
2188 ```
2189
2190 Now define the request types that MCP clients will send:
2191
2192 ```rust theme={null}
2193 #[derive(serde::Deserialize, schemars::JsonSchema)]
2194 pub struct MCPForecastRequest {
2195 latitude: f32,
2196 longitude: f32,
2197 }
2198
2199 #[derive(serde::Deserialize, schemars::JsonSchema)]
2200 pub struct MCPAlertRequest {
2201 state: String,
2202 }
2203 ```
2204
2205 ### Helper functions
2206
2207 Add helper functions for making API requests and formatting responses:
2208
2209 ```rust theme={null}
2210 async fn make_nws_request<T: DeserializeOwned>(url: &str) -> Result<T> {
2211 let client = reqwest::Client::new();
2212 let rsp = client
2213 .get(url)
2214 .header(reqwest::header::USER_AGENT, USER_AGENT)
2215 .header(reqwest::header::ACCEPT, "application/geo+json")
2216 .send()
2217 .await?
2218 .error_for_status()?;
2219 Ok(rsp.json::<T>().await?)
2220 }
2221
2222 fn format_alert(feature: &AlertFeature) -> String {
2223 let props = &feature.properties;
2224 format!(
2225 "Event: {}\nArea: {}\nSeverity: {}\nDescription: {}\nInstructions: {}",
2226 props.event.as_deref().unwrap_or("Unknown"),
2227 props.area_desc.as_deref().unwrap_or("Unknown"),
2228 props.severity.as_deref().unwrap_or("Unknown"),
2229 props
2230 .description
2231 .as_deref()
2232 .unwrap_or("No description available"),
2233 props
2234 .instruction
2235 .as_deref()
2236 .unwrap_or("No specific instructions provided")
2237 )
2238 }
2239
2240 fn format_period(period: &ForecastPeriod) -> String {
2241 format!(
2242 "{}:\nTemperature: {}°{}\nWind: {} {}\nForecast: {}",
2243 period.name,
2244 period.temperature,
2245 period.temperature_unit,
2246 period.wind_speed,
2247 period.wind_direction,
2248 period.detailed_forecast
2249 )
2250 }
2251 ```
2252
2253 ### Implementing the Weather server and tools
2254
2255 Now let's implement the main Weather server struct with the tool handlers:
2256
2257 ```rust theme={null}
2258 pub struct Weather {
2259 tool_router: ToolRouter<Weather>,
2260 }
2261
2262 #[tool_router]
2263 impl Weather {
2264 fn new() -> Self {
2265 Self {
2266 tool_router: Self::tool_router(),
2267 }
2268 }
2269
2270 #[tool(description = "Get weather alerts for a US state.")]
2271 async fn get_alerts(
2272 &self,
2273 Parameters(MCPAlertRequest { state }): Parameters<MCPAlertRequest>,
2274 ) -> String {
2275 let url = format!(
2276 "{}/alerts/active/area/{}",
2277 NWS_API_BASE,
2278 state.to_uppercase()
2279 );
2280
2281 match make_nws_request::<AlertsResponse>(&url).await {
2282 Ok(data) => {
2283 if data.features.is_empty() {
2284 "No active alerts for this state.".to_string()
2285 } else {
2286 data.features
2287 .iter()
2288 .map(format_alert)
2289 .collect::<Vec<_>>()
2290 .join("\n---\n")
2291 }
2292 }
2293 Err(_) => "Unable to fetch alerts or no alerts found.".to_string(),
2294 }
2295 }
2296
2297 #[tool(description = "Get weather forecast for a location.")]
2298 async fn get_forecast(
2299 &self,
2300 Parameters(MCPForecastRequest {
2301 latitude,
2302 longitude,
2303 }): Parameters<MCPForecastRequest>,
2304 ) -> String {
2305 let points_url = format!("{NWS_API_BASE}/points/{latitude},{longitude}");
2306 let Ok(points_data) = make_nws_request::<PointsResponse>(&points_url).await else {
2307 return "Unable to fetch forecast data for this location.".to_string();
2308 };
2309
2310 let forecast_url = points_data.properties.forecast;
2311
2312 let Ok(forecast_data) = make_nws_request::<ForecastResponse>(&forecast_url).await else {
2313 return "Unable to fetch forecast data for this location.".to_string();
2314 };
2315
2316 let periods = &forecast_data.properties.periods;
2317 let forecast_summary: String = periods
2318 .iter()
2319 .take(5) // Next 5 periods only
2320 .map(format_period)
2321 .collect::<Vec<String>>()
2322 .join("\n---\n");
2323 forecast_summary
2324 }
2325 }
2326 ```
2327
2328 The `#[tool_router]` macro automatically generates the routing logic, and the `#[tool]` attribute marks methods as MCP tools.
2329
2330 ### Implementing the ServerHandler
2331
2332 Implement the `ServerHandler` trait to define server capabilities:
2333
2334 ```rust theme={null}
2335 #[tool_handler]
2336 impl ServerHandler for Weather {
2337 fn get_info(&self) -> ServerInfo {
2338 ServerInfo {
2339 capabilities: ServerCapabilities::builder().enable_tools().build(),
2340 ..Default::default()
2341 }
2342 }
2343 }
2344 ```
2345
2346 ### Running the server
2347
2348 Finally, implement the main function to run the server with stdio transport:
2349
2350 ```rust theme={null}
2351 #[tokio::main]
2352 async fn main() -> Result<()> {
2353 let transport = (tokio::io::stdin(), tokio::io::stdout());
2354 let service = Weather::new().serve(transport).await?;
2355 service.waiting().await?;
2356 Ok(())
2357 }
2358 ```
2359
2360 Build your server with:
2361
2362 ```bash theme={null}
2363 cargo build --release
2364 ```
2365
2366 The compiled binary will be in `target/release/weather`.
2367
2368 Let's now test your server from an existing MCP host, Claude for Desktop.
2369
2370 ## Testing your server with Claude for Desktop
2371
2372 <Note>
2373 Claude for Desktop is not yet available on Linux. Linux users can proceed to the [Building a client](/docs/2024-11-05/develop/build-client) tutorial to build an MCP client that connects to the server we just built.
2374 </Note>
2375
2376 First, make sure you have Claude for Desktop installed. [You can install the latest version here.](https://claude.ai/download) If you already have Claude for Desktop, **make sure it's updated to the latest version.**
2377
2378 We'll need to configure Claude for Desktop for whichever MCP servers you want to use. To do this, open your Claude for Desktop App configuration at `~/Library/Application Support/Claude/claude_desktop_config.json` in a text editor. Make sure to create the file if it doesn't exist.
2379
2380 For example, if you have [VS Code](https://code.visualstudio.com/) installed:
2381
2382 <CodeGroup>
2383 ```bash macOS/Linux theme={null}
2384 code ~/Library/Application\ Support/Claude/claude_desktop_config.json
2385 ```
2386
2387 ```powershell Windows theme={null}
2388 code $env:AppData\Claude\claude_desktop_config.json
2389 ```
2390 </CodeGroup>
2391
2392 You'll then add your servers in the `mcpServers` key. The MCP UI elements will only show up in Claude for Desktop if at least one server is properly configured.
2393
2394 In this case, we'll add our single weather server like so:
2395
2396 <CodeGroup>
2397 ```json macOS/Linux theme={null}
2398 {
2399 "mcpServers": {
2400 "weather": {
2401 "command": "/ABSOLUTE/PATH/TO/PARENT/FOLDER/weather/target/release/weather"
2402 }
2403 }
2404 }
2405 ```
2406
2407 ```json Windows theme={null}
2408 {
2409 "mcpServers": {
2410 "weather": {
2411 "command": "C:\\ABSOLUTE\\PATH\\TO\\PARENT\\FOLDER\\weather\\target\\release\\weather.exe"
2412 }
2413 }
2414 }
2415 ```
2416 </CodeGroup>
2417
2418 <Note>
2419 Make sure you pass in the absolute path to your compiled binary. You can get this by running `pwd` on macOS/Linux or `cd` on Windows Command Prompt from your project directory. On Windows, remember to use double backslashes (`\\`) or forward slashes (`/`) in the JSON path, and add the `.exe` extension.
2420 </Note>
2421
2422 This tells Claude for Desktop:
2423
2424 1. There's an MCP server named "weather"
2425 2. Launch it by running the compiled binary at the specified path
2426
2427 Save the file, and restart **Claude for Desktop**.
2428 </Tab>
2429
2430 <Tab title="Go">
2431 Let's get started with building our weather server! [You can find the complete code for what we'll be building here.](https://github.com/modelcontextprotocol/quickstart-resources/tree/main/weather-server-go)
2432
2433 ### Prerequisite knowledge
2434
2435 This quickstart assumes you have familiarity with:
2436
2437 * Go
2438 * LLMs like Claude
2439
2440 ### Logging in MCP Servers
2441
2442 When implementing MCP servers, be careful about how you handle logging:
2443
2444 **For STDIO-based servers:** Never use `fmt.Println()` or `fmt.Printf()`, as they write to standard output (stdout). Writing to stdout will corrupt the JSON-RPC messages and break your server.
2445
2446 **For HTTP-based servers:** Standard output logging is fine since it doesn't interfere with HTTP responses.
2447
2448 ### Best Practices
2449
2450 * Use `log.Println()` (which defaults to stderr) or a logging library that writes to stderr or files.
2451 * Use `fmt.Fprintf(os.Stderr, ...)` to write to stderr explicitly.
2452
2453 ### Quick Examples
2454
2455 ```go theme={null}
2456 // ❌ Bad (STDIO)
2457 fmt.Println("Processing request")
2458
2459 // ✅ Good (STDIO)
2460 log.Println("Processing request") // defaults to stderr
2461
2462 // ✅ Good (STDIO)
2463 fmt.Fprintln(os.Stderr, "Processing request")
2464 ```
2465
2466 ### System requirements
2467
2468 * Go 1.24 or higher installed.
2469
2470 ### Set up your environment
2471
2472 First, let's install Go if you haven't already. You can download and install Go from [go.dev](https://go.dev/dl/).
2473
2474 Verify your Go installation:
2475
2476 ```bash theme={null}
2477 go version
2478 ```
2479
2480 Now, let's create and set up our project:
2481
2482 <CodeGroup>
2483 ```bash macOS/Linux theme={null}
2484 # Create a new directory for our project
2485 mkdir weather
2486 cd weather
2487
2488 # Initialize Go module
2489 go mod init weather
2490
2491 # Install dependencies
2492 go get github.com/modelcontextprotocol/go-sdk/mcp
2493
2494 # Create our server file
2495 touch main.go
2496 ```
2497
2498 ```powershell Windows theme={null}
2499 # Create a new directory for our project
2500 md weather
2501 cd weather
2502
2503 # Initialize Go module
2504 go mod init weather
2505
2506 # Install dependencies
2507 go get github.com/modelcontextprotocol/go-sdk/mcp
2508
2509 # Create our server file
2510 new-item main.go
2511 ```
2512 </CodeGroup>
2513
2514 Now let's dive into building your server.
2515
2516 ## Building your server
2517
2518 ### Importing packages and constants
2519
2520 Add these to the top of your `main.go`:
2521
2522 ```go theme={null}
2523 package main
2524
2525 import (
2526 "cmp"
2527 "context"
2528 "encoding/json"
2529 "fmt"
2530 "io"
2531 "log"
2532 "net/http"
2533 "strings"
2534
2535 "github.com/modelcontextprotocol/go-sdk/mcp"
2536 )
2537
2538 const (
2539 NWSAPIBase = "https://api.weather.gov"
2540 UserAgent = "weather-app/1.0"
2541 )
2542 ```
2543
2544 ### Data structures
2545
2546 Next, let's define the data structures used by our tools:
2547
2548 ```go theme={null}
2549 type PointsResponse struct {
2550 Properties struct {
2551 Forecast string `json:"forecast"`
2552 } `json:"properties"`
2553 }
2554
2555 type ForecastResponse struct {
2556 Properties struct {
2557 Periods []ForecastPeriod `json:"periods"`
2558 } `json:"properties"`
2559 }
2560
2561 type ForecastPeriod struct {
2562 Name string `json:"name"`
2563 Temperature int `json:"temperature"`
2564 TemperatureUnit string `json:"temperatureUnit"`
2565 WindSpeed string `json:"windSpeed"`
2566 WindDirection string `json:"windDirection"`
2567 DetailedForecast string `json:"detailedForecast"`
2568 }
2569
2570 type AlertsResponse struct {
2571 Features []AlertFeature `json:"features"`
2572 }
2573
2574 type AlertFeature struct {
2575 Properties AlertProperties `json:"properties"`
2576 }
2577
2578 type AlertProperties struct {
2579 Event string `json:"event"`
2580 AreaDesc string `json:"areaDesc"`
2581 Severity string `json:"severity"`
2582 Description string `json:"description"`
2583 Instruction string `json:"instruction"`
2584 }
2585
2586 type ForecastInput struct {
2587 Latitude float64 `json:"latitude" jsonschema:"Latitude of the location"`
2588 Longitude float64 `json:"longitude" jsonschema:"Longitude of the location"`
2589 }
2590
2591 type AlertsInput struct {
2592 State string `json:"state" jsonschema:"Two-letter US state code (e.g. CA, NY)"`
2593 }
2594 ```
2595
2596 ### Helper functions
2597
2598 Next, let's add our helper functions for querying and formatting the data from the National Weather Service API:
2599
2600 ```go theme={null}
2601 func makeNWSRequest[T any](ctx context.Context, url string) (*T, error) {
2602 req, err := http.NewRequestWithContext(ctx, http.MethodGet, url, nil)
2603 if err != nil {
2604 return nil, fmt.Errorf("failed to create request: %w", err)
2605 }
2606
2607 req.Header.Set("User-Agent", UserAgent)
2608 req.Header.Set("Accept", "application/geo+json")
2609
2610 client := http.DefaultClient
2611 resp, err := client.Do(req)
2612 if err != nil {
2613 return nil, fmt.Errorf("failed to make request to %s: %w", url, err)
2614 }
2615 defer resp.Body.Close()
2616
2617 if resp.StatusCode != http.StatusOK {
2618 body, _ := io.ReadAll(resp.Body)
2619 return nil, fmt.Errorf("HTTP error %d: %s", resp.StatusCode, string(body))
2620 }
2621
2622 var result T
2623 if err := json.NewDecoder(resp.Body).Decode(&result); err != nil {
2624 return nil, fmt.Errorf("failed to decode response: %w", err)
2625 }
2626
2627 return &result, nil
2628 }
2629
2630 func formatAlert(alert AlertFeature) string {
2631 props := alert.Properties
2632 event := cmp.Or(props.Event, "Unknown")
2633 areaDesc := cmp.Or(props.AreaDesc, "Unknown")
2634 severity := cmp.Or(props.Severity, "Unknown")
2635 description := cmp.Or(props.Description, "No description available")
2636 instruction := cmp.Or(props.Instruction, "No specific instructions provided")
2637
2638 return fmt.Sprintf(`
2639 Event: %s
2640 Area: %s
2641 Severity: %s
2642 Description: %s
2643 Instructions: %s
2644 `, event, areaDesc, severity, description, instruction)
2645 }
2646
2647 func formatPeriod(period ForecastPeriod) string {
2648 return fmt.Sprintf(`
2649 %s:
2650 Temperature: %d°%s
2651 Wind: %s %s
2652 Forecast: %s
2653 `, period.Name, period.Temperature, period.TemperatureUnit,
2654 period.WindSpeed, period.WindDirection, period.DetailedForecast)
2655 }
2656 ```
2657
2658 ### Implementing tool execution
2659
2660 The tool execution handler is responsible for actually executing the logic of each tool. Let's add it:
2661
2662 ```go theme={null}
2663 func getForecast(ctx context.Context, req *mcp.CallToolRequest, input ForecastInput) (
2664 *mcp.CallToolResult, any, error,
2665 ) {
2666 // Get points data
2667 pointsURL := fmt.Sprintf("%s/points/%f,%f", NWSAPIBase, input.Latitude, input.Longitude)
2668 pointsData, err := makeNWSRequest[PointsResponse](ctx, pointsURL)
2669 if err != nil {
2670 return &mcp.CallToolResult{
2671 Content: []mcp.Content{
2672 &mcp.TextContent{Text: "Unable to fetch forecast data for this location."},
2673 },
2674 }, nil, nil
2675 }
2676
2677 // Get forecast data
2678 forecastURL := pointsData.Properties.Forecast
2679 if forecastURL == "" {
2680 return &mcp.CallToolResult{
2681 Content: []mcp.Content{
2682 &mcp.TextContent{Text: "Unable to fetch forecast URL."},
2683 },
2684 }, nil, nil
2685 }
2686
2687 forecastData, err := makeNWSRequest[ForecastResponse](ctx, forecastURL)
2688 if err != nil {
2689 return &mcp.CallToolResult{
2690 Content: []mcp.Content{
2691 &mcp.TextContent{Text: "Unable to fetch detailed forecast."},
2692 },
2693 }, nil, nil
2694 }
2695
2696 // Format the periods
2697 periods := forecastData.Properties.Periods
2698 if len(periods) == 0 {
2699 return &mcp.CallToolResult{
2700 Content: []mcp.Content{
2701 &mcp.TextContent{Text: "No forecast periods available."},
2702 },
2703 }, nil, nil
2704 }
2705
2706 // Show next 5 periods
2707 var forecasts []string
2708 for i := range min(5, len(periods)) {
2709 forecasts = append(forecasts, formatPeriod(periods[i]))
2710 }
2711
2712 result := strings.Join(forecasts, "\n---\n")
2713
2714 return &mcp.CallToolResult{
2715 Content: []mcp.Content{
2716 &mcp.TextContent{Text: result},
2717 },
2718 }, nil, nil
2719 }
2720
2721 func getAlerts(ctx context.Context, req *mcp.CallToolRequest, input AlertsInput) (
2722 *mcp.CallToolResult, any, error,
2723 ) {
2724 // Build alerts URL
2725 stateCode := strings.ToUpper(input.State)
2726 alertsURL := fmt.Sprintf("%s/alerts/active/area/%s", NWSAPIBase, stateCode)
2727
2728 alertsData, err := makeNWSRequest[AlertsResponse](ctx, alertsURL)
2729 if err != nil {
2730 return &mcp.CallToolResult{
2731 Content: []mcp.Content{
2732 &mcp.TextContent{Text: "Unable to fetch alerts or no alerts found."},
2733 },
2734 }, nil, nil
2735 }
2736
2737 // Check if there are any alerts
2738 if len(alertsData.Features) == 0 {
2739 return &mcp.CallToolResult{
2740 Content: []mcp.Content{
2741 &mcp.TextContent{Text: "No active alerts for this state."},
2742 },
2743 }, nil, nil
2744 }
2745
2746 // Format alerts
2747 var alerts []string
2748 for _, feature := range alertsData.Features {
2749 alerts = append(alerts, formatAlert(feature))
2750 }
2751
2752 result := strings.Join(alerts, "\n---\n")
2753
2754 return &mcp.CallToolResult{
2755 Content: []mcp.Content{
2756 &mcp.TextContent{Text: result},
2757 },
2758 }, nil, nil
2759 }
2760 ```
2761
2762 ### Running the server
2763
2764 Finally, implement the main function to run the server:
2765
2766 ```go theme={null}
2767 func main() {
2768 // Create MCP server
2769 server := mcp.NewServer(&mcp.Implementation{
2770 Name: "weather",
2771 Version: "1.0.0",
2772 }, nil)
2773
2774 // Add get_forecast tool
2775 mcp.AddTool(server, &mcp.Tool{
2776 Name: "get_forecast",
2777 Description: "Get weather forecast for a location",
2778 }, getForecast)
2779
2780 // Add get_alerts tool
2781 mcp.AddTool(server, &mcp.Tool{
2782 Name: "get_alerts",
2783 Description: "Get weather alerts for a US state",
2784 }, getAlerts)
2785
2786 // Run server on stdio transport
2787 if err := server.Run(context.Background(), &mcp.StdioTransport{}); err != nil {
2788 log.Fatal(err)
2789 }
2790 }
2791 ```
2792
2793 Build your server with:
2794
2795 ```bash theme={null}
2796 go build -o weather .
2797 ```
2798
2799 The compiled binary will be in `./weather`.
2800
2801 Let's now test your server from an existing MCP host, Claude for Desktop.
2802
2803 ## Testing your server with Claude for Desktop
2804
2805 <Note>
2806 Claude for Desktop is not yet available on Linux. Linux users can proceed to the [Building a client](/docs/2024-11-05/develop/build-client) tutorial to build an MCP client that connects to the server we just built.
2807 </Note>
2808
2809 First, make sure you have Claude for Desktop installed. [You can install the latest version here.](https://claude.ai/download) If you already have Claude for Desktop, **make sure it's updated to the latest version.**
2810
2811 We'll need to configure Claude for Desktop for whichever MCP servers you want to use. To do this, open your Claude for Desktop App configuration at `~/Library/Application Support/Claude/claude_desktop_config.json` in a text editor. Make sure to create the file if it doesn't exist.
2812
2813 For example, if you have [VS Code](https://code.visualstudio.com/) installed:
2814
2815 <CodeGroup>
2816 ```bash macOS/Linux theme={null}
2817 code ~/Library/Application\ Support/Claude/claude_desktop_config.json
2818 ```
2819
2820 ```powershell Windows theme={null}
2821 code $env:AppData\Claude\claude_desktop_config.json
2822 ```
2823 </CodeGroup>
2824
2825 You'll then add your servers in the `mcpServers` key. The MCP UI elements will only show up in Claude for Desktop if at least one server is properly configured.
2826
2827 In this case, we'll add our single weather server like so:
2828
2829 <CodeGroup>
2830 ```json macOS/Linux theme={null}
2831 {
2832 "mcpServers": {
2833 "weather": {
2834 "command": "/ABSOLUTE/PATH/TO/PARENT/FOLDER/weather/weather"
2835 }
2836 }
2837 }
2838 ```
2839
2840 ```json Windows theme={null}
2841 {
2842 "mcpServers": {
2843 "weather": {
2844 "command": "C:\\ABSOLUTE\\PATH\\TO\\PARENT\\FOLDER\\weather\\weather.exe"
2845 }
2846 }
2847 }
2848 ```
2849 </CodeGroup>
2850
2851 <Note>
2852 Make sure you pass in the absolute path to your compiled binary. You can get this by running `pwd` on macOS/Linux or `cd` on Windows Command Prompt from your project directory. On Windows, remember to use double backslashes (`\\`) or forward slashes (`/`) in the JSON path, and add the `.exe` extension.
2853 </Note>
2854
2855 This tells Claude for Desktop:
2856
2857 1. There's an MCP server named "weather"
2858 2. Launch it by running the compiled binary at the specified path
2859
2860 Save the file, and restart **Claude for Desktop**.
2861 </Tab>
2862</Tabs>
2296 }
2297}
2298```
2299
2300### Running the server
2301
2302Finally, implement the main function to run the server with stdio transport:
2303
2304```rust theme={null}
2305#[tokio::main]
2306async fn main() -> Result<()> {
2307 let transport = (tokio::io::stdin(), tokio::io::stdout());
2308 let service = Weather::new().serve(transport).await?;
2309 service.waiting().await?;
2310 Ok(())
2311}
2312```
2313
2314Build your server with:
2315
2316```bash theme={null}
2317cargo build --release
2318```
2319
2320The compiled binary will be in `target/release/weather`.
2321
2322Let's now test your server from an existing MCP host, Claude for Desktop.
2323
2324## Testing your server with Claude for Desktop
2325
2326> [!NOTE]
2327>
2328> Claude for Desktop is not yet available on Linux. Linux users can proceed to the [Building a client](/docs/2024-11-05/develop/build-client) tutorial to build an MCP client that connects to the server we just built.
2329
2330First, make sure you have Claude for Desktop installed. [You can install the latest version here.](https://claude.ai/download) If you already have Claude for Desktop, **make sure it's updated to the latest version.**
2331
2332We'll need to configure Claude for Desktop for whichever MCP servers you want to use. To do this, open your Claude for Desktop App configuration at `~/Library/Application Support/Claude/claude_desktop_config.json` in a text editor. Make sure to create the file if it doesn't exist.
2333
2334For example, if you have [VS Code](https://code.visualstudio.com/) installed:
2335
2336```bash macOS/Linux theme={null}
2337code ~/Library/Application\ Support/Claude/claude_desktop_config.json
2338```
2339
2340```powershell Windows theme={null}
2341code $env:AppData\Claude\claude_desktop_config.json
2342```
2343
2344You'll then add your servers in the `mcpServers` key. The MCP UI elements will only show up in Claude for Desktop if at least one server is properly configured.
2345
2346In this case, we'll add our single weather server like so:
2347
2348```json macOS/Linux theme={null}
2349{
2350 "mcpServers": {
2351 "weather": {
2352 "command": "/ABSOLUTE/PATH/TO/PARENT/FOLDER/weather/target/release/weather"
2353 }
2354 }
2355}
2356```
2357
2358```json Windows theme={null}
2359{
2360 "mcpServers": {
2361 "weather": {
2362 "command": "C:\\ABSOLUTE\\PATH\\TO\\PARENT\\FOLDER\\weather\\target\\release\\weather.exe"
2363 }
2364 }
2365}
2366```
2367
2368> [!NOTE]
2369>
2370> Make sure you pass in the absolute path to your compiled binary. You can get this by running `pwd` on macOS/Linux or `cd` on Windows Command Prompt from your project directory. On Windows, remember to use double backslashes (`\\`) or forward slashes (`/`) in the JSON path, and add the `.exe` extension.
2371
2372This tells Claude for Desktop:
2373
23741. There's an MCP server named "weather"
23752. Launch it by running the compiled binary at the specified path
2376
2377Save the file, and restart **Claude for Desktop**.
2378
2379**Go**
2380
2381Let's get started with building our weather server! [You can find the complete code for what we'll be building here.](https://github.com/modelcontextprotocol/quickstart-resources/tree/main/weather-server-go)
2382
2383### Prerequisite knowledge
2384
2385This quickstart assumes you have familiarity with:
2386
2387* Go
2388* LLMs like Claude
2389
2390### Logging in MCP Servers
2391
2392When implementing MCP servers, be careful about how you handle logging:
2393
2394**For STDIO-based servers:** Never use `fmt.Println()` or `fmt.Printf()`, as they write to standard output (stdout). Writing to stdout will corrupt the JSON-RPC messages and break your server.
2395
2396**For HTTP-based servers:** Standard output logging is fine since it doesn't interfere with HTTP responses.
2397
2398### Best Practices
2399
2400* Use `log.Println()` (which defaults to stderr) or a logging library that writes to stderr or files.
2401* Use `fmt.Fprintf(os.Stderr, ...)` to write to stderr explicitly.
2402
2403### Quick Examples
2404
2405```go theme={null}
2406// ❌ Bad (STDIO)
2407fmt.Println("Processing request")
2408
2409// ✅ Good (STDIO)
2410log.Println("Processing request") // defaults to stderr
2411
2412// ✅ Good (STDIO)
2413fmt.Fprintln(os.Stderr, "Processing request")
2414```
2415
2416### System requirements
2417
2418* Go 1.24 or higher installed.
2419
2420### Set up your environment
2421
2422First, let's install Go if you haven't already. You can download and install Go from [go.dev](https://go.dev/dl/).
2423
2424Verify your Go installation:
2425
2426```bash theme={null}
2427go version
2428```
2429
2430Now, let's create and set up our project:
2431
2432```bash macOS/Linux theme={null}
2433# Create a new directory for our project
2434mkdir weather
2435cd weather
2436
2437# Initialize Go module
2438go mod init weather
2439
2440# Install dependencies
2441go get github.com/modelcontextprotocol/go-sdk/mcp
2442
2443# Create our server file
2444touch main.go
2445```
2446
2447```powershell Windows theme={null}
2448# Create a new directory for our project
2449md weather
2450cd weather
2451
2452# Initialize Go module
2453go mod init weather
2454
2455# Install dependencies
2456go get github.com/modelcontextprotocol/go-sdk/mcp
2457
2458# Create our server file
2459new-item main.go
2460```
2461
2462Now let's dive into building your server.
2463
2464## Building your server
2465
2466### Importing packages and constants
2467
2468Add these to the top of your `main.go`:
2469
2470```go theme={null}
2471package main
2472
2473import (
2474 "cmp"
2475 "context"
2476 "encoding/json"
2477 "fmt"
2478 "io"
2479 "log"
2480 "net/http"
2481 "strings"
2482
2483 "github.com/modelcontextprotocol/go-sdk/mcp"
2484)
2485
2486const (
2487 NWSAPIBase = "https://api.weather.gov"
2488 UserAgent = "weather-app/1.0"
2489)
2490```
2491
2492### Data structures
2493
2494Next, let's define the data structures used by our tools:
2495
2496```go theme={null}
2497type PointsResponse struct {
2498 Properties struct {
2499 Forecast string `json:"forecast"`
2500 } `json:"properties"`
2501}
2502
2503type ForecastResponse struct {
2504 Properties struct {
2505 Periods []ForecastPeriod `json:"periods"`
2506 } `json:"properties"`
2507}
2508
2509type ForecastPeriod struct {
2510 Name string `json:"name"`
2511 Temperature int `json:"temperature"`
2512 TemperatureUnit string `json:"temperatureUnit"`
2513 WindSpeed string `json:"windSpeed"`
2514 WindDirection string `json:"windDirection"`
2515 DetailedForecast string `json:"detailedForecast"`
2516}
2517
2518type AlertsResponse struct {
2519 Features []AlertFeature `json:"features"`
2520}
2521
2522type AlertFeature struct {
2523 Properties AlertProperties `json:"properties"`
2524}
2525
2526type AlertProperties struct {
2527 Event string `json:"event"`
2528 AreaDesc string `json:"areaDesc"`
2529 Severity string `json:"severity"`
2530 Description string `json:"description"`
2531 Instruction string `json:"instruction"`
2532}
2533
2534type ForecastInput struct {
2535 Latitude float64 `json:"latitude" jsonschema:"Latitude of the location"`
2536 Longitude float64 `json:"longitude" jsonschema:"Longitude of the location"`
2537}
2538
2539type AlertsInput struct {
2540 State string `json:"state" jsonschema:"Two-letter US state code (e.g. CA, NY)"`
2541}
2542```
2543
2544### Helper functions
2545
2546Next, let's add our helper functions for querying and formatting the data from the National Weather Service API:
2547
2548```go theme={null}
2549func makeNWSRequest[T any](ctx context.Context, url string) (*T, error) {
2550 req, err := http.NewRequestWithContext(ctx, http.MethodGet, url, nil)
2551 if err != nil {
2552 return nil, fmt.Errorf("failed to create request: %w", err)
2553 }
2554
2555 req.Header.Set("User-Agent", UserAgent)
2556 req.Header.Set("Accept", "application/geo+json")
2557
2558 client := http.DefaultClient
2559 resp, err := client.Do(req)
2560 if err != nil {
2561 return nil, fmt.Errorf("failed to make request to %s: %w", url, err)
2562 }
2563 defer resp.Body.Close()
2564
2565 if resp.StatusCode != http.StatusOK {
2566 body, _ := io.ReadAll(resp.Body)
2567 return nil, fmt.Errorf("HTTP error %d: %s", resp.StatusCode, string(body))
2568 }
2569
2570 var result T
2571 if err := json.NewDecoder(resp.Body).Decode(&result); err != nil {
2572 return nil, fmt.Errorf("failed to decode response: %w", err)
2573 }
2574
2575 return &result, nil
2576}
2577
2578func formatAlert(alert AlertFeature) string {
2579 props := alert.Properties
2580 event := cmp.Or(props.Event, "Unknown")
2581 areaDesc := cmp.Or(props.AreaDesc, "Unknown")
2582 severity := cmp.Or(props.Severity, "Unknown")
2583 description := cmp.Or(props.Description, "No description available")
2584 instruction := cmp.Or(props.Instruction, "No specific instructions provided")
2585
2586 return fmt.Sprintf(`
2587Event: %s
2588Area: %s
2589Severity: %s
2590Description: %s
2591Instructions: %s
2592`, event, areaDesc, severity, description, instruction)
2593}
2594
2595func formatPeriod(period ForecastPeriod) string {
2596 return fmt.Sprintf(`
2597%s:
2598Temperature: %d°%s
2599Wind: %s %s
2600Forecast: %s
2601`, period.Name, period.Temperature, period.TemperatureUnit,
2602 period.WindSpeed, period.WindDirection, period.DetailedForecast)
2603}
2604```
2605
2606### Implementing tool execution
2607
2608The tool execution handler is responsible for actually executing the logic of each tool. Let's add it:
2609
2610```go theme={null}
2611func getForecast(ctx context.Context, req *mcp.CallToolRequest, input ForecastInput) (
2612 *mcp.CallToolResult, any, error,
2613) {
2614 // Get points data
2615 pointsURL := fmt.Sprintf("%s/points/%f,%f", NWSAPIBase, input.Latitude, input.Longitude)
2616 pointsData, err := makeNWSRequest[PointsResponse](ctx, pointsURL)
2617 if err != nil {
2618 return &mcp.CallToolResult{
2619 Content: []mcp.Content{
2620 &mcp.TextContent{Text: "Unable to fetch forecast data for this location."},
2621 },
2622 }, nil, nil
2623 }
2624
2625 // Get forecast data
2626 forecastURL := pointsData.Properties.Forecast
2627 if forecastURL == "" {
2628 return &mcp.CallToolResult{
2629 Content: []mcp.Content{
2630 &mcp.TextContent{Text: "Unable to fetch forecast URL."},
2631 },
2632 }, nil, nil
2633 }
2634
2635 forecastData, err := makeNWSRequest[ForecastResponse](ctx, forecastURL)
2636 if err != nil {
2637 return &mcp.CallToolResult{
2638 Content: []mcp.Content{
2639 &mcp.TextContent{Text: "Unable to fetch detailed forecast."},
2640 },
2641 }, nil, nil
2642 }
2643
2644 // Format the periods
2645 periods := forecastData.Properties.Periods
2646 if len(periods) == 0 {
2647 return &mcp.CallToolResult{
2648 Content: []mcp.Content{
2649 &mcp.TextContent{Text: "No forecast periods available."},
2650 },
2651 }, nil, nil
2652 }
2653
2654 // Show next 5 periods
2655 var forecasts []string
2656 for i := range min(5, len(periods)) {
2657 forecasts = append(forecasts, formatPeriod(periods[i]))
2658 }
2659
2660 result := strings.Join(forecasts, "\n---\n")
2661
2662 return &mcp.CallToolResult{
2663 Content: []mcp.Content{
2664 &mcp.TextContent{Text: result},
2665 },
2666 }, nil, nil
2667}
2668
2669func getAlerts(ctx context.Context, req *mcp.CallToolRequest, input AlertsInput) (
2670 *mcp.CallToolResult, any, error,
2671) {
2672 // Build alerts URL
2673 stateCode := strings.ToUpper(input.State)
2674 alertsURL := fmt.Sprintf("%s/alerts/active/area/%s", NWSAPIBase, stateCode)
2675
2676 alertsData, err := makeNWSRequest[AlertsResponse](ctx, alertsURL)
2677 if err != nil {
2678 return &mcp.CallToolResult{
2679 Content: []mcp.Content{
2680 &mcp.TextContent{Text: "Unable to fetch alerts or no alerts found."},
2681 },
2682 }, nil, nil
2683 }
2684
2685 // Check if there are any alerts
2686 if len(alertsData.Features) == 0 {
2687 return &mcp.CallToolResult{
2688 Content: []mcp.Content{
2689 &mcp.TextContent{Text: "No active alerts for this state."},
2690 },
2691 }, nil, nil
2692 }
2693
2694 // Format alerts
2695 var alerts []string
2696 for _, feature := range alertsData.Features {
2697 alerts = append(alerts, formatAlert(feature))
2698 }
2699
2700 result := strings.Join(alerts, "\n---\n")
2701
2702 return &mcp.CallToolResult{
2703 Content: []mcp.Content{
2704 &mcp.TextContent{Text: result},
2705 },
2706 }, nil, nil
2707}
2708```
2709
2710### Running the server
2711
2712Finally, implement the main function to run the server:
2713
2714```go theme={null}
2715func main() {
2716 // Create MCP server
2717 server := mcp.NewServer(&mcp.Implementation{
2718 Name: "weather",
2719 Version: "1.0.0",
2720 }, nil)
2721
2722 // Add get_forecast tool
2723 mcp.AddTool(server, &mcp.Tool{
2724 Name: "get_forecast",
2725 Description: "Get weather forecast for a location",
2726 }, getForecast)
2727
2728 // Add get_alerts tool
2729 mcp.AddTool(server, &mcp.Tool{
2730 Name: "get_alerts",
2731 Description: "Get weather alerts for a US state",
2732 }, getAlerts)
2733
2734 // Run server on stdio transport
2735 if err := server.Run(context.Background(), &mcp.StdioTransport{}); err != nil {
2736 log.Fatal(err)
2737 }
2738}
2739```
2740
2741Build your server with:
2742
2743```bash theme={null}
2744go build -o weather .
2745```
2746
2747The compiled binary will be in `./weather`.
2748
2749Let's now test your server from an existing MCP host, Claude for Desktop.
2750
2751## Testing your server with Claude for Desktop
2752
2753> [!NOTE]
2754>
2755> Claude for Desktop is not yet available on Linux. Linux users can proceed to the [Building a client](/docs/2024-11-05/develop/build-client) tutorial to build an MCP client that connects to the server we just built.
2756
2757First, make sure you have Claude for Desktop installed. [You can install the latest version here.](https://claude.ai/download) If you already have Claude for Desktop, **make sure it's updated to the latest version.**
2758
2759We'll need to configure Claude for Desktop for whichever MCP servers you want to use. To do this, open your Claude for Desktop App configuration at `~/Library/Application Support/Claude/claude_desktop_config.json` in a text editor. Make sure to create the file if it doesn't exist.
2760
2761For example, if you have [VS Code](https://code.visualstudio.com/) installed:
2762
2763```bash macOS/Linux theme={null}
2764code ~/Library/Application\ Support/Claude/claude_desktop_config.json
2765```
2766
2767```powershell Windows theme={null}
2768code $env:AppData\Claude\claude_desktop_config.json
2769```
2770
2771You'll then add your servers in the `mcpServers` key. The MCP UI elements will only show up in Claude for Desktop if at least one server is properly configured.
2772
2773In this case, we'll add our single weather server like so:
2774
2775```json macOS/Linux theme={null}
2776{
2777 "mcpServers": {
2778 "weather": {
2779 "command": "/ABSOLUTE/PATH/TO/PARENT/FOLDER/weather/weather"
2780 }
2781 }
2782}
2783```
2784
2785```json Windows theme={null}
2786{
2787 "mcpServers": {
2788 "weather": {
2789 "command": "C:\\ABSOLUTE\\PATH\\TO\\PARENT\\FOLDER\\weather\\weather.exe"
2790 }
2791 }
2792}
2793```
2794
2795> [!NOTE]
2796>
2797> Make sure you pass in the absolute path to your compiled binary. You can get this by running `pwd` on macOS/Linux or `cd` on Windows Command Prompt from your project directory. On Windows, remember to use double backslashes (`\\`) or forward slashes (`/`) in the JSON path, and add the `.exe` extension.
2798
2799This tells Claude for Desktop:
2800
28011. There's an MCP server named "weather"
28022. Launch it by running the compiled binary at the specified path
2803
2804Save the file, and restart **Claude for Desktop**.
28632805
28642806### Test with commands
28652807
2866Let's make sure Claude for Desktop is picking up the two tools we've exposed in our `weather` server. You can do this by looking for the "Add files, connectors, and more /" <img src="https://mintcdn.com/mcp/cpXzQjUOzyH0mCNH/images/claude-add-files-connectors-and-more.png?fit=max&auto=format&n=cpXzQjUOzyH0mCNH&q=85&s=53acf21f6807dd5323b70b84b5d98d8a" style={{display: 'inline', margin: 0, height: '1.3em'}} width="33" height="33" data-path="images/claude-add-files-connectors-and-more.png" /> icon:
2867
2868<Frame>
2869 <img src="https://mintcdn.com/mcp/zNouQwo2h8cbxlDS/images/visual-indicator-mcp-tools.png?fit=max&auto=format&n=zNouQwo2h8cbxlDS&q=85&s=1bf23a2cfc5f6dd3dac1c7574cceebc9" width="684" height="133" data-path="images/visual-indicator-mcp-tools.png" />
2870</Frame>
2808Let's make sure Claude for Desktop is picking up the two tools we've exposed in our `weather` server. You can do this by looking for the "Add files, connectors, and more /"  icon:
2809
2810
28712811
28722812After clicking on the plus icon, hover over the "Connectors" menu. You should see the `weather` servers listed:
28732813
2874<Frame>
2875 <img src="https://mintcdn.com/mcp/zNouQwo2h8cbxlDS/images/available-mcp-tools.png?fit=max&auto=format&n=zNouQwo2h8cbxlDS&q=85&s=e2ace1ac88895a5fe30ebd8d01456bc3" width="437" height="244" data-path="images/available-mcp-tools.png" />
2876</Frame>
2814
28772815
28782816If your server isn't being picked up by Claude for Desktop, proceed to the [Troubleshooting](#troubleshooting) section for debugging tips.
28792817
from line 2820
28822820* What's the weather in Sacramento?
28832821* What are the active weather alerts in Texas?
28842822
2885<Frame>
2886 <img src="https://mintcdn.com/mcp/4ZXF1PrDkEaJvXpn/images/current-weather.png?fit=max&auto=format&n=4ZXF1PrDkEaJvXpn&q=85&s=dce7b2f8a06c20ba358e4bd2e75fa4c7" width="2780" height="1849" data-path="images/current-weather.png" />
2887</Frame>
2888
2889<Frame>
2890 <img src="https://mintcdn.com/mcp/4ZXF1PrDkEaJvXpn/images/weather-alerts.png?fit=max&auto=format&n=4ZXF1PrDkEaJvXpn&q=85&s=c4762bf2bd84a8781846d2965af3e4a4" width="2809" height="1850" data-path="images/weather-alerts.png" />
2891</Frame>
2892
2893<Note>
2894 Since this is the US National Weather service, the queries will only work for US locations.
2895</Note>
2823
2824
2825
2826
2827> [!NOTE]
2828>
2829> Since this is the US National Weather service, the queries will only work for US locations.
28962830
28972831## What's happening under the hood
28982832
from line 2841
29072841
29082842## Troubleshooting
29092843
2910<AccordionGroup>
2911 <Accordion title="Claude for Desktop Integration Issues">
2912 **Getting logs from Claude for Desktop**
2913
2914 Claude.app logging related to MCP is written to log files in `~/Library/Logs/Claude`:
2915
2916 * `mcp.log` will contain general logging about MCP connections and connection failures.
2917 * Files named `mcp-server-SERVERNAME.log` will contain error (stderr) logging from the named server.
2918
2919 You can run the following command to list recent logs and follow along with any new ones:
2920
2921 ```bash theme={null}
2922 # Check Claude's logs for errors
2923 tail -n 20 -f ~/Library/Logs/Claude/mcp*.log
2924 ```
2925
2926 **Server not showing up in Claude**
2927
2928 1. Check your `claude_desktop_config.json` file syntax
2929 2. Make sure the path to your project is absolute and not relative
2930 3. Restart Claude for Desktop completely
2931
2932 <Warning>
2933 To properly restart Claude for Desktop, you must fully quit the application:
2934
2935 * **Windows**: Right-click the Claude icon in the system tray (which may be hidden in the "hidden icons" menu) and select "Quit" or "Exit".
2936 * **macOS**: Use Cmd+Q or select "Quit Claude" from the menu bar.
2937
2938 Simply closing the window does not fully quit the application, and your MCP server configuration changes will not take effect.
2939 </Warning>
2940
2941 **Tool calls failing silently**
2942
2943 If Claude attempts to use the tools but they fail:
2944
2945 1. Check Claude's logs for errors
2946 2. Verify your server builds and runs without errors
2947 3. Try restarting Claude for Desktop
2948
2949 **None of this is working. What do I do?**
2950
2951 Please refer to our [debugging guide](/docs/2024-11-05/tools/debugging) for better debugging tools and more detailed guidance.
2952 </Accordion>
2953
2954 <Accordion title="Weather API Issues">
2955 **Error: Failed to retrieve grid point data**
2956
2957 This usually means either:
2958
2959 1. The coordinates are outside the US
2960 2. The NWS API is having issues
2961 3. You're being rate limited
2962
2963 Fix:
2964
2965 * Verify you're using US coordinates
2966 * Add a small delay between requests
2967 * Check the NWS API status page
2968
2969 **Error: No active alerts for \[STATE]**
2970
2971 This isn't an error - it just means there are no current weather alerts for that state. Try a different state or check during severe weather.
2972 </Accordion>
2973</AccordionGroup>
2974
2975<Note>
2976 For more advanced troubleshooting, check out our guide on [Debugging MCP](/docs/2024-11-05/tools/debugging)
2977</Note>
2844> **Claude for Desktop Integration Issues**
2845>
2846> **Getting logs from Claude for Desktop**
2847>
2848> Claude.app logging related to MCP is written to log files in `~/Library/Logs/Claude`:
2849>
2850> * `mcp.log` will contain general logging about MCP connections and connection failures.
2851> * Files named `mcp-server-SERVERNAME.log` will contain error (stderr) logging from the named server.
2852>
2853> You can run the following command to list recent logs and follow along with any new ones:
2854>
2855> ```bash theme={null}
2856> # Check Claude's logs for errors
2857> tail -n 20 -f ~/Library/Logs/Claude/mcp*.log
2858> ```
2859>
2860> **Server not showing up in Claude**
2861>
2862> 1. Check your `claude_desktop_config.json` file syntax
2863>
2864> 2. Make sure the path to your project is absolute and not relative
2865>
2866> 3. Restart Claude for Desktop completely
2867>
2868> > [!WARNING]
2869> >
2870> > To properly restart Claude for Desktop, you must fully quit the application:
2871> >
2872> > * **Windows**: Right-click the Claude icon in the system tray (which may be hidden in the "hidden icons" menu) and select "Quit" or "Exit".
2873> > * **macOS**: Use Cmd+Q or select "Quit Claude" from the menu bar.
2874> >
2875> > Simply closing the window does not fully quit the application, and your MCP server configuration changes will not take effect.
2876>
2877> **Tool calls failing silently**
2878>
2879> If Claude attempts to use the tools but they fail:
2880>
2881> 1. Check Claude's logs for errors
2882> 2. Verify your server builds and runs without errors
2883> 3. Try restarting Claude for Desktop
2884>
2885> **None of this is working. What do I do?**
2886>
2887> Please refer to our [debugging guide](/docs/2024-11-05/tools/debugging) for better debugging tools and more detailed guidance.
2888
2889> **Weather API Issues**
2890>
2891> **Error: Failed to retrieve grid point data**
2892>
2893> This usually means either:
2894>
2895> 1. The coordinates are outside the US
2896> 2. The NWS API is having issues
2897> 3. You're being rate limited
2898>
2899> Fix:
2900>
2901> * Verify you're using US coordinates
2902> * Add a small delay between requests
2903> * Check the NWS API status page
2904>
2905> **Error: No active alerts for \[STATE]**
2906>
2907> This isn't an error - it just means there are no current weather alerts for that state. Try a different state or check during severe weather.
2908
2909> [!NOTE]
2910>
2911> For more advanced troubleshooting, check out our guide on [Debugging MCP](/docs/2024-11-05/tools/debugging)
29782912
29792913## Next steps
29802914
2981<CardGroup cols={2}>
2982 <Card title="Building a client" icon="outlet" href="/docs/2024-11-05/develop/build-client">
2983 Learn how to build your own MCP client that can connect to your server
2984 </Card>
2985
2986 <Card title="Example servers" icon="grid" href="/examples">
2987 Check out our gallery of official MCP servers and implementations
2988 </Card>
2989
2990 <Card title="Debugging Guide" icon="bug" href="/docs/2024-11-05/tools/debugging">
2991 Learn how to effectively debug MCP servers and integrations
2992 </Card>
2993
2994 <Card title="Build with Agent Skills" icon="comments" href="/docs/2024-11-05/develop/build-with-agent-skills">
2995 Use agent skills to guide AI coding assistants through server design
2996 </Card>
2997</CardGroup>
2915> **[Building a client](/docs/2024-11-05/develop/build-client)**
2916>
2917> Learn how to build your own MCP client that can connect to your server
2918
2919> **[Example servers](/examples)**
2920>
2921> Check out our gallery of official MCP servers and implementations
2922
2923> **[Debugging Guide](/docs/2024-11-05/tools/debugging)**
2924>
2925> Learn how to effectively debug MCP servers and integrations
2926
2927> **[Build with Agent Skills](/docs/2024-11-05/develop/build-with-agent-skills)**
2928>
2929> Use agent skills to guide AI coding assistants through server design
29982930
docs/2024-11-05/develop/build-with-agent-skills Changed · +9 / -11 lines
from line 83
8383Once your agent scaffolds the server, iterate on tool descriptions and error
8484handling, then test and ship:
8585
86<CardGroup cols={2}>
87 <Card title="MCP Inspector" icon="magnifying-glass" href="/docs/2024-11-05/tools/inspector">
88 Test your server's tools, resources, and prompts interactively
89 </Card>
86> **[MCP Inspector](/docs/2024-11-05/tools/inspector)**
87>
88> Test your server's tools, resources, and prompts interactively
9089
91 <Card title="Connect to a client" icon="plug" href="/docs/2024-11-05/develop/connect-local-servers">
92 Wire your server into an MCP client via local or remote configuration
93 </Card>
90> **[Connect to a client](/docs/2024-11-05/develop/connect-local-servers)**
91>
92> Wire your server into an MCP client via local or remote configuration
9493
95 <Card title="Publish to the Registry" icon="box" href="/registry/quickstart">
96 Make your server discoverable in the MCP Registry
97 </Card>
98</CardGroup>
94> **[Publish to the Registry](/registry/quickstart)**
95>
96> Make your server discoverable in the MCP Registry
9997
docs/2024-11-05/develop/connect-local-servers Changed · +188 / -212 lines
from line 6
66
77This guide demonstrates how to connect to local MCP servers using Claude Desktop as an example, one of the many clients that support MCP. While we focus on Claude Desktop's implementation, the concepts apply broadly to other MCP-compatible clients. By the end of this tutorial, Claude will be able to interact with files on your computer, create new documents, organize folders, and search through your file system—all with your explicit permission for each action.
88
9<Frame>
10 <img src="https://mintcdn.com/mcp/4ZXF1PrDkEaJvXpn/images/quickstart-filesystem.png?fit=max&auto=format&n=4ZXF1PrDkEaJvXpn&q=85&s=629d7e754dc358d71a408d6ce970c1b1" alt="Claude Desktop with filesystem integration showing file management capabilities" width="1732" height="2060" data-path="images/quickstart-filesystem.png" />
11</Frame>
9
1210
1311## Prerequisites
1412
from line 43
4543
4644The process involves configuring Claude Desktop to automatically start the Filesystem Server whenever you launch the application. This configuration is done through a JSON file that tells Claude Desktop which servers to run and how to connect to them.
4745
48<Steps>
49 <Step title="Open Claude Desktop Settings">
50 Start by accessing the Claude Desktop settings. Click on the Claude menu in your system's menu bar (not the settings within the Claude window itself) and select "Settings..."
46**1. Open Claude Desktop Settings**
5147
52 On macOS, this appears in the top menu bar:
48Start by accessing the Claude Desktop settings. Click on the Claude menu in your system's menu bar (not the settings within the Claude window itself) and select "Settings..."
5349
54 <Frame style={{ textAlign: "center" }}>
55 <img src="https://mintcdn.com/mcp/4ZXF1PrDkEaJvXpn/images/quickstart-menu.png?fit=max&auto=format&n=4ZXF1PrDkEaJvXpn&q=85&s=0c8b57e0e17af3624b6762a3ea944c8e" width="400" alt="Claude Desktop menu showing Settings option" data-path="images/quickstart-menu.png" />
56 </Frame>
50On macOS, this appears in the top menu bar:
5751
58 This opens the Claude Desktop configuration window, which is separate from your Claude account settings.
59 </Step>
52
6053
61 <Step title="Access Developer Settings">
62 In the Settings window, navigate to the "Developer" tab in the left sidebar. This section contains options for configuring MCP servers and other developer features.
54This opens the Claude Desktop configuration window, which is separate from your Claude account settings.
6355
64 Click the "Edit Config" button to open the configuration file:
56**2. Access Developer Settings**
6557
66 <Frame>
67 <img src="https://mintcdn.com/mcp/4ZXF1PrDkEaJvXpn/images/quickstart-developer.png?fit=max&auto=format&n=4ZXF1PrDkEaJvXpn&q=85&s=0fb595490a2f9e15c0301e771a57446c" alt="Developer settings showing Edit Config button" width="1688" height="534" data-path="images/quickstart-developer.png" />
68 </Frame>
58In the Settings window, navigate to the "Developer" tab in the left sidebar. This section contains options for configuring MCP servers and other developer features.
6959
70 This action creates a new configuration file if one doesn't exist, or opens your existing configuration. The file is located at:
60Click the "Edit Config" button to open the configuration file:
7161
72 * **macOS**: `~/Library/Application Support/Claude/claude_desktop_config.json`
73 * **Windows**: `%APPDATA%\Claude\claude_desktop_config.json`
74 </Step>
62
7563
76 <Step title="Configure the Filesystem Server">
77 Replace the contents of the configuration file with the following JSON structure. This configuration tells Claude Desktop to start the Filesystem Server with access to specific directories:
64This action creates a new configuration file if one doesn't exist, or opens your existing configuration. The file is located at:
7865
79 <CodeGroup>
80 ```json macOS theme={null}
81 {
82 "mcpServers": {
83 "filesystem": {
84 "command": "npx",
85 "args": [
86 "-y",
87 "@modelcontextprotocol/server-filesystem",
88 "/Users/username/Desktop",
89 "/Users/username/Downloads"
90 ]
91 }
92 }
93 }
94 ```
66* **macOS**: `~/Library/Application Support/Claude/claude_desktop_config.json`
67* **Windows**: `%APPDATA%\Claude\claude_desktop_config.json`
9568
96 ```json Windows theme={null}
97 {
98 "mcpServers": {
99 "filesystem": {
100 "command": "npx",
101 "args": [
102 "-y",
103 "@modelcontextprotocol/server-filesystem",
104 "C:\\Users\\username\\Desktop",
105 "C:\\Users\\username\\Downloads"
106 ]
107 }
108 }
109 }
110 ```
111 </CodeGroup>
69**3. Configure the Filesystem Server**
11270
113 Replace `username` with your actual computer username. The paths listed in the `args` array specify which directories the Filesystem Server can access. You can modify these paths or add additional directories as needed.
71Replace the contents of the configuration file with the following JSON structure. This configuration tells Claude Desktop to start the Filesystem Server with access to specific directories:
11472
115 <Tip>
116 **Understanding the Configuration**
73```json macOS theme={null}
74{
75 "mcpServers": {
76 "filesystem": {
77 "command": "npx",
78 "args": [
79 "-y",
80 "@modelcontextprotocol/server-filesystem",
81 "/Users/username/Desktop",
82 "/Users/username/Downloads"
83 ]
84 }
85 }
86}
87```
11788
118 * `"filesystem"`: A friendly name for the server that appears in Claude Desktop
119 * `"command": "npx"`: Uses Node.js's npx tool to run the server
120 * `"-y"`: Automatically confirms the installation of the server package
121 * `"@modelcontextprotocol/server-filesystem"`: The package name of the Filesystem Server
122 * The remaining arguments: Directories the server is allowed to access
123 </Tip>
89```json Windows theme={null}
90{
91 "mcpServers": {
92 "filesystem": {
93 "command": "npx",
94 "args": [
95 "-y",
96 "@modelcontextprotocol/server-filesystem",
97 "C:\\Users\\username\\Desktop",
98 "C:\\Users\\username\\Downloads"
99 ]
100 }
101 }
102}
103```
124104
125 <Warning>
126 **Security Consideration**
105Replace `username` with your actual computer username. The paths listed in the `args` array specify which directories the Filesystem Server can access. You can modify these paths or add additional directories as needed.
127106
128 Only grant access to directories you're comfortable with Claude reading and modifying. The server runs with your user account permissions, so it can perform any file operations you can perform manually.
129 </Warning>
130 </Step>
107> [!TIP]
108>
109> **Understanding the Configuration**
110>
111> * `"filesystem"`: A friendly name for the server that appears in Claude Desktop
112> * `"command": "npx"`: Uses Node.js's npx tool to run the server
113> * `"-y"`: Automatically confirms the installation of the server package
114> * `"@modelcontextprotocol/server-filesystem"`: The package name of the Filesystem Server
115> * The remaining arguments: Directories the server is allowed to access
131116
132 <Step title="Restart Claude Desktop">
133 After saving the configuration file, completely quit Claude Desktop and restart it. The application needs to restart to load the new configuration and start the MCP server.
117> [!WARNING]
118>
119> **Security Consideration**
120>
121> Only grant access to directories you're comfortable with Claude reading and modifying. The server runs with your user account permissions, so it can perform any file operations you can perform manually.
134122
135 Upon successful restart, click the "Add files, connectors and more" indicator <img src="https://mintcdn.com/mcp/cpXzQjUOzyH0mCNH/images/claude-add-files-connectors-and-more.png?fit=max&auto=format&n=cpXzQjUOzyH0mCNH&q=85&s=53acf21f6807dd5323b70b84b5d98d8a" style={{display: 'inline', margin: 0, height: '1.3em'}} width="33" height="33" data-path="images/claude-add-files-connectors-and-more.png" /> in the bottom-left corner of the conversation input box:
123**4. Restart Claude Desktop**
136124
137 <Frame>
138 <img src="https://mintcdn.com/mcp/akpggzunDlIcY2im/images/quickstart-slider.png?fit=max&auto=format&n=akpggzunDlIcY2im&q=85&s=a1ebd4259cff2a7472171885f2edc035" alt="Claude Desktop interface showing MCP server indicator" width="1414" height="410" data-path="images/quickstart-slider.png" />
139 </Frame>
125After saving the configuration file, completely quit Claude Desktop and restart it. The application needs to restart to load the new configuration and start the MCP server.
140126
141 Click on this indicator, then scroll over "Connectors" and click "Manage connectors". Select "filesystem" from the connector list to view the the Filesystem Server's available tools:
127Upon successful restart, click the "Add files, connectors and more" indicator  in the bottom-left corner of the conversation input box:
142128
143 <Frame style={{ textAlign: "center" }}>
144 <img src="https://mintcdn.com/mcp/cpXzQjUOzyH0mCNH/images/quickstart-tools.png?fit=max&auto=format&n=cpXzQjUOzyH0mCNH&q=85&s=212a63d76daba170d52db0d2f6f582be" width="400" alt="Available filesystem tools in Claude Desktop" data-path="images/quickstart-tools.png" />
145 </Frame>
129
146130
147 If the Filesystem Server doesn't connect, refer to the [Troubleshooting](#troubleshooting) section for debugging steps.
148 </Step>
149</Steps>
131Click on this indicator, then scroll over "Connectors" and click "Manage connectors". Select "filesystem" from the connector list to view the the Filesystem Server's available tools:
150132
133
134
135If the Filesystem Server doesn't connect, refer to the [Troubleshooting](#troubleshooting) section for debugging steps.
136
151137## Using the Filesystem Server
152138
153139With the Filesystem Server connected, Claude can now interact with your file system. Try these example requests to explore the capabilities:
from line 148
162148
163149Before executing any file system operation, Claude will request your approval. This ensures you maintain control over all actions:
164150
165<Frame style={{ textAlign: "center" }}>
166 <img src="https://mintcdn.com/mcp/4ZXF1PrDkEaJvXpn/images/quickstart-approve.png?fit=max&auto=format&n=4ZXF1PrDkEaJvXpn&q=85&s=98cc6e9dfe885fbd6e9bfae40601e494" width="500" alt="Claude requesting approval to perform a file operation" data-path="images/quickstart-approve.png" />
167</Frame>
151
168152
169153Review each request carefully before approving. You can always deny a request if you're not comfortable with the proposed action.
170154
from line 156
172156
173157If you encounter issues setting up or using the Filesystem Server, these solutions address common problems:
174158
175<AccordionGroup>
176 <Accordion title="Server not showing up in Claude / hammer icon missing">
177 1. Restart Claude Desktop completely
178 2. Check your `claude_desktop_config.json` file syntax
179 3. Make sure the file paths included in `claude_desktop_config.json` are valid and that they are absolute and not relative
180 4. Look at [logs](#getting-logs-from-claude-for-desktop) to see why the server is not connecting
181 5. In your command line, try manually running the server (replacing `username` as you did in `claude_desktop_config.json`) to see if you get any errors:
159> **Server not showing up in Claude / hammer icon missing**
160>
161> 1. Restart Claude Desktop completely
162> 2. Check your `claude_desktop_config.json` file syntax
163> 3. Make sure the file paths included in `claude_desktop_config.json` are valid and that they are absolute and not relative
164> 4. Look at [logs](#getting-logs-from-claude-for-desktop) to see why the server is not connecting
165> 5. In your command line, try manually running the server (replacing `username` as you did in `claude_desktop_config.json`) to see if you get any errors:
166>
167> ```bash macOS/Linux theme={null}
168> npx -y @modelcontextprotocol/server-filesystem /Users/username/Desktop /Users/username/Downloads
169> ```
170>
171> ```powershell Windows theme={null}
172> npx -y @modelcontextprotocol/server-filesystem C:\Users\username\Desktop C:\Users\username\Downloads
173> ```
182174
183 <CodeGroup>
184 ```bash macOS/Linux theme={null}
185 npx -y @modelcontextprotocol/server-filesystem /Users/username/Desktop /Users/username/Downloads
186 ```
175> **Getting logs from Claude Desktop**
176>
177> Claude.app logging related to MCP is written to log files in:
178>
179> * macOS: `~/Library/Logs/Claude`
180>
181> * Windows: `%APPDATA%\Claude\logs`
182>
183> * `mcp.log` will contain general logging about MCP connections and connection failures.
184>
185> * Files named `mcp-server-SERVERNAME.log` will contain error (stderr) logging from the named server.
186>
187> You can run the following command to list recent logs and follow along with any new ones (on Windows, it will only show recent logs):
188>
189> ```bash macOS/Linux theme={null}
190> tail -n 20 -f ~/Library/Logs/Claude/mcp*.log
191> ```
192>
193> ```powershell Windows theme={null}
194> type "%APPDATA%\Claude\logs\mcp*.log"
195> ```
187196
188 ```powershell Windows theme={null}
189 npx -y @modelcontextprotocol/server-filesystem C:\Users\username\Desktop C:\Users\username\Downloads
190 ```
191 </CodeGroup>
192 </Accordion>
197> **Tool calls failing silently**
198>
199> If Claude attempts to use the tools but they fail:
200>
201> 1. Check Claude's logs for errors
202> 2. Verify your server builds and runs without errors
203> 3. Try restarting Claude Desktop
193204
194 <Accordion title="Getting logs from Claude Desktop">
195 Claude.app logging related to MCP is written to log files in:
205> **None of this is working. What do I do?**
206>
207> Please refer to our [debugging guide](/docs/2024-11-05/tools/debugging) for better debugging tools and more detailed guidance.
196208
197 * macOS: `~/Library/Logs/Claude`
209> **ENOENT error and `$\{APPDATA\}` in paths on Windows**
210>
211> If your configured server fails to load, and you see within its logs an error referring to `${APPDATA}` within a path, you may need to add the expanded value of `%APPDATA%` to your `env` key in `claude_desktop_config.json`:
212>
213> ```json theme={null}
214> {
215> "brave-search": {
216> "command": "npx",
217> "args": ["-y", "@modelcontextprotocol/server-brave-search"],
218> "env": {
219> "APPDATA": "C:\\Users\\user\\AppData\\Roaming\\",
220> "BRAVE_API_KEY": "..."
221> }
222> }
223> }
224> ```
225>
226> With this change in place, launch Claude Desktop once again.
227>
228> > [!WARNING]
229> >
230> > **npm should be installed globally**
231> >
232> > The `npx` command may continue to fail if you have not installed npm globally. If npm is already installed globally, you will find `%APPDATA%\npm` exists on your system. If not, you can install npm globally by running the following command:
233> >
234> > ```bash theme={null}
235> > npm install -g npm
236> > ```
198237
199 * Windows: `%APPDATA%\Claude\logs`
200
201 * `mcp.log` will contain general logging about MCP connections and connection failures.
202
203 * Files named `mcp-server-SERVERNAME.log` will contain error (stderr) logging from the named server.
204
205 You can run the following command to list recent logs and follow along with any new ones (on Windows, it will only show recent logs):
206
207 <CodeGroup>
208 ```bash macOS/Linux theme={null}
209 tail -n 20 -f ~/Library/Logs/Claude/mcp*.log
210 ```
211
212 ```powershell Windows theme={null}
213 type "%APPDATA%\Claude\logs\mcp*.log"
214 ```
215 </CodeGroup>
216 </Accordion>
217
218 <Accordion title="Tool calls failing silently">
219 If Claude attempts to use the tools but they fail:
220
221 1. Check Claude's logs for errors
222 2. Verify your server builds and runs without errors
223 3. Try restarting Claude Desktop
224 </Accordion>
225
226 <Accordion title="None of this is working. What do I do?">
227 Please refer to our [debugging guide](/docs/2024-11-05/tools/debugging) for better debugging tools and more detailed guidance.
228 </Accordion>
229
230 <Accordion title="ENOENT error and `${APPDATA}` in paths on Windows">
231 If your configured server fails to load, and you see within its logs an error referring to `${APPDATA}` within a path, you may need to add the expanded value of `%APPDATA%` to your `env` key in `claude_desktop_config.json`:
232
233 ```json theme={null}
234 {
235 "brave-search": {
236 "command": "npx",
237 "args": ["-y", "@modelcontextprotocol/server-brave-search"],
238 "env": {
239 "APPDATA": "C:\\Users\\user\\AppData\\Roaming\\",
240 "BRAVE_API_KEY": "..."
241 }
242 }
243 }
244 ```
245
246 With this change in place, launch Claude Desktop once again.
247
248 <Warning>
249 **npm should be installed globally**
250
251 The `npx` command may continue to fail if you have not installed npm globally. If npm is already installed globally, you will find `%APPDATA%\npm` exists on your system. If not, you can install npm globally by running the following command:
252
253 ```bash theme={null}
254 npm install -g npm
255 ```
256 </Warning>
257 </Accordion>
258</AccordionGroup>
259
260238## Next Steps
261239
262240Now that you've successfully connected Claude Desktop to a local MCP server, explore these options to expand your setup:
263241
264<CardGroup cols={2}>
265 <Card title="Explore other servers" icon="grid" href="https://github.com/modelcontextprotocol/servers">
266 Browse our collection of official and community-created MCP servers for
267 additional capabilities
268 </Card>
242> **[Explore other servers](https://github.com/modelcontextprotocol/servers)**
243>
244> Browse our collection of official and community-created MCP servers for
245> additional capabilities
269246
270 <Card title="Build your own server" icon="code" href="/docs/2024-11-05/develop/build-server">
271 Create custom MCP servers tailored to your specific workflows and
272 integrations
273 </Card>
247> **[Build your own server](/docs/2024-11-05/develop/build-server)**
248>
249> Create custom MCP servers tailored to your specific workflows and
250> integrations
274251
275 <Card title="Connect to remote servers" icon="cloud" href="/docs/2024-11-05/develop/connect-remote-servers">
276 Learn how to connect Claude to remote MCP servers for cloud-based tools and
277 services
278 </Card>
252> **[Connect to remote servers](/docs/2024-11-05/develop/connect-remote-servers)**
253>
254> Learn how to connect Claude to remote MCP servers for cloud-based tools and
255> services
279256
280 <Card title="Understand the protocol" icon="book" href="/docs/2024-11-05/learn/architecture">
281 Dive deeper into how MCP works and its architecture
282 </Card>
283</CardGroup>
257> **[Understand the protocol](/docs/2024-11-05/learn/architecture)**
258>
259> Dive deeper into how MCP works and its architecture
284260
docs/2024-11-05/develop/connect-remote-servers Changed · +41 / -57 lines
from line 25
2525
2626The process of connecting Claude to a remote MCP server involves adding a Custom Connector through the [Claude interface](https://claude.ai/). This establishes a secure connection between Claude and your chosen remote server.
2727
28<Steps>
29 <Step title="Navigate to Connector Settings">
30 Open Claude in your browser and navigate to the settings page. You can access this by clicking on your profile icon and selecting "Settings" from the dropdown menu. Once in settings, locate and click on the "Connectors" section in the sidebar.
28**1. Navigate to Connector Settings**
3129
32 This will display your currently configured connectors and provide options to add new ones.
33 </Step>
30Open Claude in your browser and navigate to the settings page. You can access this by clicking on your profile icon and selecting "Settings" from the dropdown menu. Once in settings, locate and click on the "Connectors" section in the sidebar.
3431
35 <Step title="Add a Custom Connector">
36 In the Connectors section, scroll to the bottom where you'll find the "Add custom connector" button. Click this button to begin the connection process.
32This will display your currently configured connectors and provide options to add new ones.
3733
38 <Frame>
39 <img src="https://mintcdn.com/mcp/4ZXF1PrDkEaJvXpn/images/quickstart-remote/1-add-connector.png?fit=max&auto=format&n=4ZXF1PrDkEaJvXpn&q=85&s=b5ae9b23164875bbaa3aff4c178cdc64" alt="Add custom connector button in Claude settings" width="1038" height="809" data-path="images/quickstart-remote/1-add-connector.png" />
40 </Frame>
34**2. Add a Custom Connector**
4135
42 A dialog will appear prompting you to enter the remote MCP server URL. This URL should be provided by the server developer or administrator. Enter the complete URL, ensuring it includes the proper protocol (https\://) and any necessary path components.
36In the Connectors section, scroll to the bottom where you'll find the "Add custom connector" button. Click this button to begin the connection process.
4337
44 <Frame>
45 <img src="https://mintcdn.com/mcp/4ZXF1PrDkEaJvXpn/images/quickstart-remote/2-connect.png?fit=max&auto=format&n=4ZXF1PrDkEaJvXpn&q=85&s=0934f16d8e016cade8e560c8f89d011b" alt="Dialog for entering remote MCP server URL" width="1616" height="282" data-path="images/quickstart-remote/2-connect.png" />
46 </Frame>
38
4739
48 After entering the URL, click "Add" to proceed with the connection.
49 </Step>
40A dialog will appear prompting you to enter the remote MCP server URL. This URL should be provided by the server developer or administrator. Enter the complete URL, ensuring it includes the proper protocol (https\://) and any necessary path components.
5041
51 <Step title="Complete Authentication">
52 Most remote MCP servers require authentication to ensure secure access to their resources. The authentication process varies depending on the server implementation but commonly involves OAuth, API keys, or username/password combinations.
42
5343
54 <Frame>
55 <img src="https://mintcdn.com/mcp/4ZXF1PrDkEaJvXpn/images/quickstart-remote/3-auth.png?fit=max&auto=format&n=4ZXF1PrDkEaJvXpn&q=85&s=89af6e1b85718637231388697cc7b015" alt="Authentication screen for remote MCP server" width="490" height="806" data-path="images/quickstart-remote/3-auth.png" />
56 </Frame>
44After entering the URL, click "Add" to proceed with the connection.
5745
58 Follow the authentication prompts provided by the server. This may redirect you to a third-party authentication provider or display a form within Claude. Once authentication is complete, Claude will establish a secure connection to the remote server.
59 </Step>
46**3. Complete Authentication**
6047
61 <Step title="Access Resources and Prompts">
62 After successful connection, the remote server's resources and prompts become available in your Claude conversations. You can access these by clicking the paperclip icon in the message input area, which opens the attachment menu.
48Most remote MCP servers require authentication to ensure secure access to their resources. The authentication process varies depending on the server implementation but commonly involves OAuth, API keys, or username/password combinations.
6349
64 <Frame>
65 <img src="https://mintcdn.com/mcp/cpXzQjUOzyH0mCNH/images/quickstart-remote/4-select-resources-menu.png?fit=max&auto=format&n=cpXzQjUOzyH0mCNH&q=85&s=e5fa85174f8acbadbd709bac66f42d5c" alt="Attachment menu showing available resources" width="735" height="378" data-path="images/quickstart-remote/4-select-resources-menu.png" />
66 </Frame>
50
6751
68 The menu displays all available resources and prompts from your connected servers. Select the items you want to include in your conversation. These resources provide Claude with context and information from your external tools.
52Follow the authentication prompts provided by the server. This may redirect you to a third-party authentication provider or display a form within Claude. Once authentication is complete, Claude will establish a secure connection to the remote server.
6953
70 <Frame>
71 <img src="https://mintcdn.com/mcp/cpXzQjUOzyH0mCNH/images/quickstart-remote/5-select-prompts-resources.png?fit=max&auto=format&n=cpXzQjUOzyH0mCNH&q=85&s=68722669d9e18252756885c703e4f221" alt="Selecting specific resources and prompts from the menu" width="648" height="529" data-path="images/quickstart-remote/5-select-prompts-resources.png" />
72 </Frame>
73 </Step>
54**4. Access Resources and Prompts**
7455
75 <Step title="Configure Tool Permissions">
76 Remote MCP servers often expose multiple tools with varying capabilities. You can control which tools Claude is allowed to use by configuring permissions in the connector settings. This ensures Claude only performs actions you've explicitly authorized.
56After successful connection, the remote server's resources and prompts become available in your Claude conversations. You can access these by clicking the paperclip icon in the message input area, which opens the attachment menu.
7757
78 <Frame>
79 <img src="https://mintcdn.com/mcp/cpXzQjUOzyH0mCNH/images/quickstart-remote/6-configure-tools.png?fit=max&auto=format&n=cpXzQjUOzyH0mCNH&q=85&s=5cfd8b2c5d06e7e3699eac24c68d090e" alt="Tool permission configuration interface" width="604" height="745" data-path="images/quickstart-remote/6-configure-tools.png" />
80 </Frame>
58
8159
82 Navigate back to the Connectors settings and click on your connected server. Here you can enable or disable specific tools, set usage limits, and configure other security parameters according to your needs.
83 </Step>
84</Steps>
60The menu displays all available resources and prompts from your connected servers. Select the items you want to include in your conversation. These resources provide Claude with context and information from your external tools.
8561
62
63
64**5. Configure Tool Permissions**
65
66Remote MCP servers often expose multiple tools with varying capabilities. You can control which tools Claude is allowed to use by configuring permissions in the connector settings. This ensures Claude only performs actions you've explicitly authorized.
67
68
69
70Navigate back to the Connectors settings and click on your connected server. Here you can enable or disable specific tools, set usage limits, and configure other security parameters according to your needs.
71
8672## Best Practices for Using Remote MCP Servers
8773
8874When working with remote MCP servers, consider these recommendations to ensure a secure and efficient experience:
from line 81
9581
9682Now that you've connected Claude to a remote MCP server, you can explore its capabilities in your conversations. Try using the connected tools to automate tasks, access external data, or integrate with your existing workflows.
9783
98<CardGroup cols={2}>
99 <Card title="Build your own remote server" icon="cloud" href="/docs/2024-11-05/develop/build-server">
100 Create custom remote MCP servers to integrate with proprietary tools and
101 services
102 </Card>
84> **[Build your own remote server](/docs/2024-11-05/develop/build-server)**
85>
86> Create custom remote MCP servers to integrate with proprietary tools and
87> services
10388
104 <Card title="Explore available servers" icon="grid" href="https://github.com/modelcontextprotocol/servers">
105 Browse our collection of official and community-created MCP servers
106 </Card>
89> **[Explore available servers](https://github.com/modelcontextprotocol/servers)**
90>
91> Browse our collection of official and community-created MCP servers
10792
108 <Card title="Connect local servers" icon="computer" href="/docs/2024-11-05/develop/connect-local-servers">
109 Learn how to connect Claude Desktop to local MCP servers for direct system
110 access
111 </Card>
93> **[Connect local servers](/docs/2024-11-05/develop/connect-local-servers)**
94>
95> Learn how to connect Claude Desktop to local MCP servers for direct system
96> access
11297
113 <Card title="Understand the architecture" icon="book" href="/docs/2024-11-05/learn/architecture">
114 Dive deeper into how MCP works and its architecture
115 </Card>
116</CardGroup>
98> **[Understand the architecture](/docs/2024-11-05/learn/architecture)**
99>
100> Dive deeper into how MCP works and its architecture
117101
118102Remote MCP servers unlock powerful possibilities for extending Claude's capabilities. As you become familiar with these integrations, you'll discover new ways to streamline your workflows and accomplish complex tasks more efficiently.
119103
docs/2024-11-05/getting-started/intro Changed · +19 / -27 lines
from line 6
66
77Think of MCP like a USB-C port for AI applications. Just as USB-C provides a standardized way to connect electronic devices, MCP provides a standardized way to connect AI applications to external systems.
88
9<Frame>
10 <img src="https://mintcdn.com/mcp/bEUxYpZqie0DsluH/images/mcp-simple-diagram.png?fit=max&auto=format&n=bEUxYpZqie0DsluH&q=85&s=35268aa0ad50b8c385913810e7604550" width="3840" height="1500" data-path="images/mcp-simple-diagram.png" />
11</Frame>
9
1210
1311## What can MCP enable?
1412
from line 29
3129
3230## Start Building
3331
34<CardGroup cols={2}>
35 <Card title="Build servers" icon="server" href="/docs/2024-11-05/develop/build-server">
36 Create MCP servers to expose your data and tools
37 </Card>
32> **[Build servers](/docs/2024-11-05/develop/build-server)**
33>
34> Create MCP servers to expose your data and tools
3835
39 <Card title="Build clients" icon="computer" href="/docs/2024-11-05/develop/build-client">
40 Develop applications that connect to MCP servers
41 </Card>
36> **[Build clients](/docs/2024-11-05/develop/build-client)**
37>
38> Develop applications that connect to MCP servers
4239
43 <Card title="Build MCP Apps" icon="puzzle-piece" href="/extensions/apps/overview">
44 Build interactive apps that run inside AI clients
45 </Card>
46</CardGroup>
40> **[Build MCP Apps](/extensions/apps/overview)**
41>
42> Build interactive apps that run inside AI clients
4743
4844## Learn more
4945
50<CardGroup cols={2}>
51 <Card title="Architecture" icon="book" href="/docs/2024-11-05/learn/architecture">
52 Learn the core concepts and architecture of MCP
53 </Card>
46> **[Architecture](/docs/2024-11-05/learn/architecture)**
47>
48> Learn the core concepts and architecture of MCP
5449
55 <Card title="Security" icon="lock" href="/docs/2024-11-05/tutorials/security/security_best_practices">
56 Understand the security considerations and best practices for MCP
57 </Card>
58</CardGroup>
50> **[Security](/docs/2024-11-05/tutorials/security/security_best_practices)**
51>
52> Understand the security considerations and best practices for MCP
5953
6054## Community
6155
62<CardGroup cols={2}>
63 <Card title="Contributing" icon="users" href="/community/contributing">
64 Learn how to get involved and contribute to MCP
65 </Card>
66</CardGroup>
56> **[Contributing](/community/contributing)**
57>
58> Learn how to get involved and contribute to MCP
6759
docs/2024-11-05/learn/architecture Changed · +304 / -312 lines
#### Understanding the Initialization Exchange #### How This Works in AI Applications #### Understanding the Tool Discovery Request #### Understanding the Tool Discovery Response #### How This Works in AI Applications #### Understanding the Tool Execution Request #### Key Elements of Tool Execution #### Understanding the Tool Execution Response #### How This Works in AI Applications #### Understanding Tool List Change Notifications #### Key Features of MCP Notifications #### Client Response to Notifications #### Why Notifications Matter #### How This Works in AI Applications
The two sides of this change are more than 400 edits apart, too far apart to line up, so this is the differ's own diff of it and the words inside a line are not marked.
from line 15
1515* **MCP Development Tools**: Tools for developing MCP servers and clients, including the [MCP Inspector](https://github.com/modelcontextprotocol/inspector)
1616* [MCP Reference Server Implementations](https://github.com/modelcontextprotocol/servers): Reference implementations of MCP servers.
1717
18<Note>
19 MCP focuses solely on the protocol for context exchange—it does not dictate
20 how AI applications use LLMs or manage the provided context.
21</Note>
18> [!NOTE]
19>
20> MCP focuses solely on the protocol for context exchange—it does not dictate
21> how AI applications use LLMs or manage the provided context.
2222
2323## Concepts of MCP
2424
from line 104
104104
105105#### Lifecycle management
106106
107MCP is a stateful protocol that requires lifecycle management. The purpose of lifecycle management is to negotiate the <Tooltip tip="Features and operations that a client or server supports, such as tools, resources, or prompts">capabilities</Tooltip> that both client and server support. Detailed information can be found in the [specification](/specification/2024-11-05/basic/lifecycle), and the [example](#example) showcases the initialization sequence.
107MCP is a stateful protocol that requires lifecycle management. The purpose of lifecycle management is to negotiate the capabilities (Features and operations that a client or server supports, such as tools, resources, or prompts) that both client and server support. Detailed information can be found in the [specification](/specification/2024-11-05/basic/lifecycle), and the [example](#example) showcases the initialization sequence.
108108
109109#### Primitives
110110
from line 144
144144
145145This section provides a step-by-step walkthrough of an MCP client-server interaction, focusing on the data layer protocol. We'll demonstrate the lifecycle sequence, tool operations, and notifications using JSON-RPC 2.0 messages.
146146
147<Steps>
148 <Step title="Initialization (Lifecycle Management)">
149 MCP begins with lifecycle management through a capability negotiation handshake. As described in the [lifecycle management](#lifecycle-management) section, the client sends an `initialize` request to establish the connection and negotiate supported features.
150
151 <CodeGroup>
152 ```json Initialize Request theme={null}
147**1. Initialization (Lifecycle Management)**
148
149MCP begins with lifecycle management through a capability negotiation handshake. As described in the [lifecycle management](#lifecycle-management) section, the client sends an `initialize` request to establish the connection and negotiate supported features.
150
151```json Initialize Request theme={null}
152{
153 "jsonrpc": "2.0",
154 "id": 1,
155 "method": "initialize",
156 "params": {
157 "protocolVersion": "2024-11-05",
158 "capabilities": {
159 "sampling": {}
160 },
161 "clientInfo": {
162 "name": "example-client",
163 "version": "1.0.0"
164 }
165 }
166}
167```
168
169```json Initialize Response theme={null}
170{
171 "jsonrpc": "2.0",
172 "id": 1,
173 "result": {
174 "protocolVersion": "2024-11-05",
175 "capabilities": {
176 "tools": {
177 "listChanged": true
178 },
179 "resources": {}
180 },
181 "serverInfo": {
182 "name": "example-server",
183 "version": "1.0.0"
184 }
185 }
186}
187```
188
189#### Understanding the Initialization Exchange
190
191The initialization process is a key part of MCP's lifecycle management and serves several critical purposes:
192
1931. **Protocol Version Negotiation**: The `protocolVersion` field (e.g., "2024-11-05") ensures both client and server are using compatible protocol versions. This prevents communication errors that could occur when different versions attempt to interact. If a mutually compatible version is not negotiated, the connection should be terminated.
194
1952. **Capability Discovery**: The `capabilities` object allows each party to declare what features they support, including which [primitives](#primitives) they can handle (tools, resources, prompts) and whether they support features like [notifications](#notifications). This enables efficient communication by avoiding unsupported operations.
196
1973. **Identity Exchange**: The `clientInfo` and `serverInfo` objects provide identification and versioning information for debugging and compatibility purposes.
198
199In this example, the capability negotiation demonstrates how MCP primitives are declared:
200
201**Client Capabilities**:
202
203* `"sampling": {}` - The client declares it can handle server sampling requests (can receive `sampling/createMessage` method calls)
204
205**Server Capabilities**:
206
207* `"tools": {"listChanged": true}` - The server supports the tools primitive AND can send `tools/list_changed` notifications when its tool list changes
208* `"resources": {}` - The server also supports the resources primitive (can handle `resources/list` and `resources/read` methods)
209
210After successful initialization, the client sends a notification to indicate it's ready:
211
212```json Notification theme={null}
213{
214 "jsonrpc": "2.0",
215 "method": "notifications/initialized"
216}
217```
218
219#### How This Works in AI Applications
220
221During initialization, the AI application's MCP client manager establishes connections to configured servers and stores their capabilities for later use. The application uses this information to determine which servers can provide specific types of functionality (tools, resources, prompts) and whether they support real-time updates.
222
223```python Pseudo-code for AI application initialization theme={null}
224# Pseudo Code
225async with stdio_client(server_config) as (read, write):
226 async with ClientSession(read, write) as session:
227 init_response = await session.initialize()
228 if init_response.capabilities.tools:
229 app.register_mcp_server(session, supports_tools=True)
230 app.set_server_ready(session)
231```
232
233**2. Tool Discovery (Primitives)**
234
235Now that the connection is established, the client can discover available tools by sending a `tools/list` request. This request is fundamental to MCP's tool discovery mechanism — it allows clients to understand what tools are available on the server before attempting to use them.
236
237```json Tools List Request theme={null}
238{
239 "jsonrpc": "2.0",
240 "id": 2,
241 "method": "tools/list"
242}
243```
244
245```json Tools List Response theme={null}
246{
247 "jsonrpc": "2.0",
248 "id": 2,
249 "result": {
250 "tools": [
153251 {
154 "jsonrpc": "2.0",
155 "id": 1,
156 "method": "initialize",
157 "params": {
158 "protocolVersion": "2024-11-05",
159 "capabilities": {
160 "sampling": {}
252 "name": "calculator_arithmetic",
253 "title": "Calculator",
254 "description": "Perform mathematical calculations including basic arithmetic, trigonometric functions, and algebraic operations",
255 "inputSchema": {
256 "type": "object",
257 "properties": {
258 "expression": {
259 "type": "string",
260 "description": "Mathematical expression to evaluate (e.g., '2 + 3 * 4', 'sin(30)', 'sqrt(16)')"
261 }
161262 },
162 "clientInfo": {
163 "name": "example-client",
164 "version": "1.0.0"
165 }
263 "required": ["expression"]
264 }
265 },
266 {
267 "name": "weather_current",
268 "title": "Weather Information",
269 "description": "Get current weather information for any location worldwide",
270 "inputSchema": {
271 "type": "object",
272 "properties": {
273 "location": {
274 "type": "string",
275 "description": "City name, address, or coordinates (latitude,longitude)"
276 },
277 "units": {
278 "type": "string",
279 "enum": ["metric", "imperial", "kelvin"],
280 "description": "Temperature units to use in response",
281 "default": "metric"
282 }
283 },
284 "required": ["location"]
166285 }
167286 }
168 ```
169
170 ```json Initialize Response theme={null}
287 ]
288 }
289}
290```
291
292#### Understanding the Tool Discovery Request
293
294The `tools/list` request is simple, containing no parameters.
295
296#### Understanding the Tool Discovery Response
297
298The response contains a `tools` array that provides comprehensive metadata about each available tool. This array-based structure allows servers to expose multiple tools simultaneously while maintaining clear boundaries between different functionalities.
299
300Each tool object in the response includes several key fields:
301
302* **`name`**: A unique identifier for the tool within the server's namespace. This serves as the primary key for tool execution and should follow a clear naming pattern (e.g., `calculator_arithmetic` rather than just `calculate`)
303* **`title`**: A human-readable display name for the tool that clients can show to users
304* **`description`**: Detailed explanation of what the tool does and when to use it
305* **`inputSchema`**: A JSON Schema that defines the expected input parameters, enabling type validation and providing clear documentation about required and optional parameters
306
307#### How This Works in AI Applications
308
309The AI application fetches available tools from all connected MCP servers and combines them into a unified tool registry that the language model can access. This allows the LLM to understand what actions it can perform and automatically generates the appropriate tool calls during conversations.
310
311```python Pseudo-code for AI application tool discovery theme={null}
312# Pseudo-code using MCP Python SDK patterns
313available_tools = []
314for session in app.mcp_server_sessions():
315 tools_response = await session.list_tools()
316 available_tools.extend(tools_response.tools)
317conversation.register_available_tools(available_tools)
318```
319
320**3. Tool Execution (Primitives)**
321
322The client can now execute a tool using the `tools/call` method. This demonstrates how MCP primitives are used in practice: after discovering available tools, the client can invoke them with appropriate arguments.
323
324#### Understanding the Tool Execution Request
325
326The `tools/call` request follows a structured format that ensures type safety and clear communication between client and server. Note that we're using the proper tool name from the discovery response (`weather_current`) rather than a simplified name:
327
328```json Tool Call Request theme={null}
329{
330 "jsonrpc": "2.0",
331 "id": 3,
332 "method": "tools/call",
333 "params": {
334 "name": "weather_current",
335 "arguments": {
336 "location": "San Francisco",
337 "units": "imperial"
338 }
339 }
340}
341```
342
343```json Tool Call Response theme={null}
344{
345 "jsonrpc": "2.0",
346 "id": 3,
347 "result": {
348 "content": [
171349 {
172 "jsonrpc": "2.0",
173 "id": 1,
174 "result": {
175 "protocolVersion": "2024-11-05",
176 "capabilities": {
177 "tools": {
178 "listChanged": true
179 },
180 "resources": {}
181 },
182 "serverInfo": {
183 "name": "example-server",
184 "version": "1.0.0"
185 }
186 }
350 "type": "text",
351 "text": "Current weather in San Francisco: 68°F, partly cloudy with light winds from the west at 8 mph. Humidity: 65%"
187352 }
188 ```
189 </CodeGroup>
190
191 #### Understanding the Initialization Exchange
192
193 The initialization process is a key part of MCP's lifecycle management and serves several critical purposes:
194
195 1. **Protocol Version Negotiation**: The `protocolVersion` field (e.g., "2024-11-05") ensures both client and server are using compatible protocol versions. This prevents communication errors that could occur when different versions attempt to interact. If a mutually compatible version is not negotiated, the connection should be terminated.
196
197 2. **Capability Discovery**: The `capabilities` object allows each party to declare what features they support, including which [primitives](#primitives) they can handle (tools, resources, prompts) and whether they support features like [notifications](#notifications). This enables efficient communication by avoiding unsupported operations.
198
199 3. **Identity Exchange**: The `clientInfo` and `serverInfo` objects provide identification and versioning information for debugging and compatibility purposes.
200
201 In this example, the capability negotiation demonstrates how MCP primitives are declared:
202
203 **Client Capabilities**:
204
205 * `"sampling": {}` - The client declares it can handle server sampling requests (can receive `sampling/createMessage` method calls)
206
207 **Server Capabilities**:
208
209 * `"tools": {"listChanged": true}` - The server supports the tools primitive AND can send `tools/list_changed` notifications when its tool list changes
210 * `"resources": {}` - The server also supports the resources primitive (can handle `resources/list` and `resources/read` methods)
211
212 After successful initialization, the client sends a notification to indicate it's ready:
213
214 ```json Notification theme={null}
215 {
216 "jsonrpc": "2.0",
217 "method": "notifications/initialized"
218 }
219 ```
220
221 #### How This Works in AI Applications
222
223 During initialization, the AI application's MCP client manager establishes connections to configured servers and stores their capabilities for later use. The application uses this information to determine which servers can provide specific types of functionality (tools, resources, prompts) and whether they support real-time updates.
224
225 ```python Pseudo-code for AI application initialization theme={null}
226 # Pseudo Code
227 async with stdio_client(server_config) as (read, write):
228 async with ClientSession(read, write) as session:
229 init_response = await session.initialize()
230 if init_response.capabilities.tools:
231 app.register_mcp_server(session, supports_tools=True)
232 app.set_server_ready(session)
233 ```
234 </Step>
235
236 <Step title="Tool Discovery (Primitives)">
237 Now that the connection is established, the client can discover available tools by sending a `tools/list` request. This request is fundamental to MCP's tool discovery mechanism — it allows clients to understand what tools are available on the server before attempting to use them.
238
239 <CodeGroup>
240 ```json Tools List Request theme={null}
241 {
242 "jsonrpc": "2.0",
243 "id": 2,
244 "method": "tools/list"
245 }
246 ```
247
248 ```json Tools List Response theme={null}
249 {
250 "jsonrpc": "2.0",
251 "id": 2,
252 "result": {
253 "tools": [
254 {
255 "name": "calculator_arithmetic",
256 "title": "Calculator",
257 "description": "Perform mathematical calculations including basic arithmetic, trigonometric functions, and algebraic operations",
258 "inputSchema": {
259 "type": "object",
260 "properties": {
261 "expression": {
262 "type": "string",
263 "description": "Mathematical expression to evaluate (e.g., '2 + 3 * 4', 'sin(30)', 'sqrt(16)')"
264 }
265 },
266 "required": ["expression"]
267 }
268 },
269 {
270 "name": "weather_current",
271 "title": "Weather Information",
272 "description": "Get current weather information for any location worldwide",
273 "inputSchema": {
274 "type": "object",
275 "properties": {
276 "location": {
277 "type": "string",
278 "description": "City name, address, or coordinates (latitude,longitude)"
279 },
280 "units": {
281 "type": "string",
282 "enum": ["metric", "imperial", "kelvin"],
283 "description": "Temperature units to use in response",
284 "default": "metric"
285 }
286 },
287 "required": ["location"]
288 }
289 }
290 ]
291 }
292 }
293 ```
294 </CodeGroup>
295
296 #### Understanding the Tool Discovery Request
297
298 The `tools/list` request is simple, containing no parameters.
299
300 #### Understanding the Tool Discovery Response
301
302 The response contains a `tools` array that provides comprehensive metadata about each available tool. This array-based structure allows servers to expose multiple tools simultaneously while maintaining clear boundaries between different functionalities.
303
304 Each tool object in the response includes several key fields:
305
306 * **`name`**: A unique identifier for the tool within the server's namespace. This serves as the primary key for tool execution and should follow a clear naming pattern (e.g., `calculator_arithmetic` rather than just `calculate`)
307 * **`title`**: A human-readable display name for the tool that clients can show to users
308 * **`description`**: Detailed explanation of what the tool does and when to use it
309 * **`inputSchema`**: A JSON Schema that defines the expected input parameters, enabling type validation and providing clear documentation about required and optional parameters
310
311 #### How This Works in AI Applications
312
313 The AI application fetches available tools from all connected MCP servers and combines them into a unified tool registry that the language model can access. This allows the LLM to understand what actions it can perform and automatically generates the appropriate tool calls during conversations.
314
315 ```python Pseudo-code for AI application tool discovery theme={null}
316 # Pseudo-code using MCP Python SDK patterns
317 available_tools = []
318 for session in app.mcp_server_sessions():
319 tools_response = await session.list_tools()
320 available_tools.extend(tools_response.tools)
321 conversation.register_available_tools(available_tools)
322 ```
323 </Step>
324
325 <Step title="Tool Execution (Primitives)">
326 The client can now execute a tool using the `tools/call` method. This demonstrates how MCP primitives are used in practice: after discovering available tools, the client can invoke them with appropriate arguments.
327
328 #### Understanding the Tool Execution Request
329
330 The `tools/call` request follows a structured format that ensures type safety and clear communication between client and server. Note that we're using the proper tool name from the discovery response (`weather_current`) rather than a simplified name:
331
332 <CodeGroup>
333 ```json Tool Call Request theme={null}
334 {
335 "jsonrpc": "2.0",
336 "id": 3,
337 "method": "tools/call",
338 "params": {
339 "name": "weather_current",
340 "arguments": {
341 "location": "San Francisco",
342 "units": "imperial"
343 }
344 }
345 }
346 ```
347
348 ```json Tool Call Response theme={null}
349 {
350 "jsonrpc": "2.0",
351 "id": 3,
352 "result": {
353 "content": [
354 {
355 "type": "text",
356 "text": "Current weather in San Francisco: 68°F, partly cloudy with light winds from the west at 8 mph. Humidity: 65%"
357 }
358 ]
359 }
360 }
361 ```
362 </CodeGroup>
363
364 #### Key Elements of Tool Execution
365
366 The request structure includes several important components:
367
368 1. **`name`**: Must match exactly the tool name from the discovery response (`weather_current`). This ensures the server can correctly identify which tool to execute.
369
370 2. **`arguments`**: Contains the input parameters as defined by the tool's `inputSchema`. In this example:
371 * `location`: "San Francisco" (required parameter)
372 * `units`: "imperial" (optional parameter, defaults to "metric" if not specified)
373
374 3. **JSON-RPC Structure**: Uses standard JSON-RPC 2.0 format with unique `id` for request-response correlation.
375
376 #### Understanding the Tool Execution Response
377
378 The response demonstrates MCP's flexible content system:
379
380 1. **`content` Array**: Tool responses return an array of content objects, allowing for rich, multi-format responses (text, images, resources, etc.)
381
382 2. **Content Types**: Each content object has a `type` field. In this example, `"type": "text"` indicates plain text content, but MCP supports various content types for different use cases.
383
384 3. **Structured Output**: The response provides actionable information that the AI application can use as context for language model interactions.
385
386 This execution pattern allows AI applications to dynamically invoke server functionality and receive structured responses that can be integrated into conversations with language models.
387
388 #### How This Works in AI Applications
389
390 When the language model decides to use a tool during a conversation, the AI application intercepts the tool call, routes it to the appropriate MCP server, executes it, and returns the results back to the LLM as part of the conversation flow. This enables the LLM to access real-time data and perform actions in the external world.
391
392 ```python theme={null}
393 # Pseudo-code for AI application tool execution
394 async def handle_tool_call(conversation, tool_name, arguments):
395 session = app.find_mcp_session_for_tool(tool_name)
396 result = await session.call_tool(tool_name, arguments)
397 conversation.add_tool_result(result.content)
398 ```
399 </Step>
400
401 <Step title="Real-time Updates (Notifications)">
402 MCP supports real-time notifications that enable servers to inform clients about changes without being explicitly requested. This demonstrates the notification system, a key feature that keeps MCP connections synchronized and responsive.
403
404 #### Understanding Tool List Change Notifications
405
406 When the server's available tools change—such as when new functionality becomes available, existing tools are modified, or tools become temporarily unavailable—the server can proactively notify connected clients:
407
408 ```json Request theme={null}
409 {
410 "jsonrpc": "2.0",
411 "method": "notifications/tools/list_changed"
412 }
413 ```
414
415 #### Key Features of MCP Notifications
416
417 1. **No Response Required**: Notice there's no `id` field in the notification. This follows JSON-RPC 2.0 notification semantics where no response is expected or sent.
418
419 2. **Capability-Based**: This notification is only sent by servers that declared `"listChanged": true` in their tools capability during initialization (as shown in Step 1).
420
421 3. **Event-Driven**: The server decides when to send notifications based on internal state changes, making MCP connections dynamic and responsive.
422
423 #### Client Response to Notifications
424
425 Upon receiving this notification, the client typically reacts by requesting the updated tool list. This creates a refresh cycle that keeps the client's understanding of available tools current:
426
427 ```json Request theme={null}
428 {
429 "jsonrpc": "2.0",
430 "id": 4,
431 "method": "tools/list"
432 }
433 ```
434
435 #### Why Notifications Matter
436
437 This notification system is crucial for several reasons:
438
439 1. **Dynamic Environments**: Tools may come and go based on server state, external dependencies, or user permissions
440 2. **Efficiency**: Clients don't need to poll for changes; they're notified when updates occur
441 3. **Consistency**: Ensures clients always have accurate information about available server capabilities
442 4. **Real-time Collaboration**: Enables responsive AI applications that can adapt to changing contexts
443
444 This notification pattern extends beyond tools to other MCP primitives, enabling comprehensive real-time synchronization between clients and servers.
445
446 #### How This Works in AI Applications
447
448 When the AI application receives a notification about changed tools, it immediately refreshes its tool registry and updates the LLM's available capabilities. This ensures that ongoing conversations always have access to the most current set of tools, and the LLM can dynamically adapt to new functionality as it becomes available.
449
450 ```python theme={null}
451 # Pseudo-code for AI application notification handling
452 async def handle_tools_changed_notification(session):
453 tools_response = await session.list_tools()
454 app.update_available_tools(session, tools_response.tools)
455 if app.conversation.is_active():
456 app.conversation.notify_llm_of_new_capabilities()
457 ```
458 </Step>
459</Steps>
353 ]
354 }
355}
356```
357
358#### Key Elements of Tool Execution
359
360The request structure includes several important components:
361
3621. **`name`**: Must match exactly the tool name from the discovery response (`weather_current`). This ensures the server can correctly identify which tool to execute.
363
3642. **`arguments`**: Contains the input parameters as defined by the tool's `inputSchema`. In this example:
365 * `location`: "San Francisco" (required parameter)
366 * `units`: "imperial" (optional parameter, defaults to "metric" if not specified)
367
3683. **JSON-RPC Structure**: Uses standard JSON-RPC 2.0 format with unique `id` for request-response correlation.
369
370#### Understanding the Tool Execution Response
371
372The response demonstrates MCP's flexible content system:
373
3741. **`content` Array**: Tool responses return an array of content objects, allowing for rich, multi-format responses (text, images, resources, etc.)
375
3762. **Content Types**: Each content object has a `type` field. In this example, `"type": "text"` indicates plain text content, but MCP supports various content types for different use cases.
377
3783. **Structured Output**: The response provides actionable information that the AI application can use as context for language model interactions.
379
380This execution pattern allows AI applications to dynamically invoke server functionality and receive structured responses that can be integrated into conversations with language models.
381
382#### How This Works in AI Applications
383
384When the language model decides to use a tool during a conversation, the AI application intercepts the tool call, routes it to the appropriate MCP server, executes it, and returns the results back to the LLM as part of the conversation flow. This enables the LLM to access real-time data and perform actions in the external world.
385
386```python theme={null}
387# Pseudo-code for AI application tool execution
388async def handle_tool_call(conversation, tool_name, arguments):
389 session = app.find_mcp_session_for_tool(tool_name)
390 result = await session.call_tool(tool_name, arguments)
391 conversation.add_tool_result(result.content)
392```
393
394**4. Real-time Updates (Notifications)**
395
396MCP supports real-time notifications that enable servers to inform clients about changes without being explicitly requested. This demonstrates the notification system, a key feature that keeps MCP connections synchronized and responsive.
397
398#### Understanding Tool List Change Notifications
399
400When the server's available tools change—such as when new functionality becomes available, existing tools are modified, or tools become temporarily unavailable—the server can proactively notify connected clients:
401
402```json Request theme={null}
403{
404 "jsonrpc": "2.0",
405 "method": "notifications/tools/list_changed"
406}
407```
408
409#### Key Features of MCP Notifications
410
4111. **No Response Required**: Notice there's no `id` field in the notification. This follows JSON-RPC 2.0 notification semantics where no response is expected or sent.
412
4132. **Capability-Based**: This notification is only sent by servers that declared `"listChanged": true` in their tools capability during initialization (as shown in Step 1).
414
4153. **Event-Driven**: The server decides when to send notifications based on internal state changes, making MCP connections dynamic and responsive.
416
417#### Client Response to Notifications
418
419Upon receiving this notification, the client typically reacts by requesting the updated tool list. This creates a refresh cycle that keeps the client's understanding of available tools current:
420
421```json Request theme={null}
422{
423 "jsonrpc": "2.0",
424 "id": 4,
425 "method": "tools/list"
426}
427```
428
429#### Why Notifications Matter
430
431This notification system is crucial for several reasons:
432
4331. **Dynamic Environments**: Tools may come and go based on server state, external dependencies, or user permissions
4342. **Efficiency**: Clients don't need to poll for changes; they're notified when updates occur
4353. **Consistency**: Ensures clients always have accurate information about available server capabilities
4364. **Real-time Collaboration**: Enables responsive AI applications that can adapt to changing contexts
437
438This notification pattern extends beyond tools to other MCP primitives, enabling comprehensive real-time synchronization between clients and servers.
439
440#### How This Works in AI Applications
441
442When the AI application receives a notification about changed tools, it immediately refreshes its tool registry and updates the LLM's available capabilities. This ensures that ongoing conversations always have access to the most current set of tools, and the LLM can dynamically adapt to new functionality as it becomes available.
443
444```python theme={null}
445# Pseudo-code for AI application notification handling
446async def handle_tools_changed_notification(session):
447 tools_response = await session.list_tools()
448 app.update_available_tools(session, tools_response.tools)
449 if app.conversation.is_active():
450 app.conversation.notify_llm_of_new_capabilities()
451```
460452
docs/2024-11-05/learn/server-concepts Changed · +3 / -3 lines
from line 10
1010
1111| Feature | Explanation | Examples | Who controls it |
1212| - | - | - | - |
13| **Tools** | Functions that your LLM can actively call, and decides when to use them based on user requests. Tools can write to databases, call external APIs, modify files, or trigger other logic. | Search flights<br />Send messages<br />Create calendar events | Model |
14| **Resources** | Passive data sources that provide read-only access to information for context, such as file contents, database schemas, or API documentation. | Retrieve documents<br />Access knowledge bases<br />Read calendars | Application |
15| **Prompts** | Pre-built instruction templates that tell the model to work with specific tools and resources. | Plan a vacation<br />Summarize my meetings<br />Draft an email | User |
13| **Tools** | Functions that your LLM can actively call, and decides when to use them based on user requests. Tools can write to databases, call external APIs, modify files, or trigger other logic. | Search flights Send messages Create calendar events | Model |
14| **Resources** | Passive data sources that provide read-only access to information for context, such as file contents, database schemas, or API documentation. | Retrieve documents Access knowledge bases Read calendars | Application |
15| **Prompts** | Pre-built instruction templates that tell the model to work with specific tools and resources. | Plan a vacation Summarize my meetings Draft an email | User |
1616
1717We will use a hypothetical scenario to demonstrate the role of each of these features, and show how they can work together.
1818
docs/2024-11-05/learn/versioning Changed · +5 / -5 lines
from line 3
33The Model Context Protocol uses string-based version identifiers following the format
44`YYYY-MM-DD`, to indicate the last date backwards incompatible changes were made.
55
6<Info>
7 The protocol version will *not* be incremented when the
8 protocol is updated, as long as the changes maintain backwards compatibility. This allows
9 for incremental improvements while preserving interoperability.
10</Info>
6> [!NOTE]
7>
8> The protocol version will *not* be incremented when the
9> protocol is updated, as long as the changes maintain backwards compatibility. This allows
10> for incremental improvements while preserving interoperability.
1111
1212## Revisions
1313
docs/2024-11-05/sdk Changed · +16 / -18 lines
from line 8
88
99| SDK | Repository | Tier |
1010| :- | :- | -: |
11| <Icon icon="square-js" size={24} /> [TypeScript](https://ts.sdk.modelcontextprotocol.io) | [modelcontextprotocol/typescript-sdk](https://github.com/modelcontextprotocol/typescript-sdk) | <Badge color="blue" shape="pill">Tier 1</Badge> |
12| <Icon icon="python" size={24} /> [Python](https://py.sdk.modelcontextprotocol.io) | [modelcontextprotocol/python-sdk](https://github.com/modelcontextprotocol/python-sdk) | <Badge color="blue" shape="pill">Tier 1</Badge> |
13| <Icon icon="square-c" size={24} /> [C#](https://csharp.sdk.modelcontextprotocol.io) | [modelcontextprotocol/csharp-sdk](https://github.com/modelcontextprotocol/csharp-sdk) | <Badge color="blue" shape="pill">Tier 1</Badge> |
14| <Icon icon="golang" size={24} /> [Go](https://go.sdk.modelcontextprotocol.io) | [modelcontextprotocol/go-sdk](https://github.com/modelcontextprotocol/go-sdk) | <Badge color="blue" shape="pill">Tier 1</Badge> |
15| <Icon icon="java" size={24} /> [Java](https://java.sdk.modelcontextprotocol.io) | [modelcontextprotocol/java-sdk](https://github.com/modelcontextprotocol/java-sdk) | <Badge color="purple" shape="pill">Tier 2</Badge> |
16| <Icon icon="rust" size={24} /> [Rust](https://rust.sdk.modelcontextprotocol.io) | [modelcontextprotocol/rust-sdk](https://github.com/modelcontextprotocol/rust-sdk) | <Badge color="purple" shape="pill">Tier 2</Badge> |
17| <Icon icon="swift" size={24} /> Swift | [modelcontextprotocol/swift-sdk](https://github.com/modelcontextprotocol/swift-sdk) | <Badge color="orange" shape="pill">Tier 3</Badge> |
18| <Icon icon="gem" size={24} /> [Ruby](https://ruby.sdk.modelcontextprotocol.io) | [modelcontextprotocol/ruby-sdk](https://github.com/modelcontextprotocol/ruby-sdk) | <Badge color="orange" shape="pill">Tier 3</Badge> |
19| <Icon icon="php" size={24} /> [PHP](https://php.sdk.modelcontextprotocol.io) | [modelcontextprotocol/php-sdk](https://github.com/modelcontextprotocol/php-sdk) | <Badge color="orange" shape="pill">Tier 3</Badge> |
20| <Icon icon="square-k" size={24} /> [Kotlin](https://kotlin.sdk.modelcontextprotocol.io) | [modelcontextprotocol/kotlin-sdk](https://github.com/modelcontextprotocol/kotlin-sdk) | <Badge color="orange" shape="pill">Tier 3</Badge> |
11| [TypeScript](https://ts.sdk.modelcontextprotocol.io) | [modelcontextprotocol/typescript-sdk](https://github.com/modelcontextprotocol/typescript-sdk) | Tier 1 |
12| [Python](https://py.sdk.modelcontextprotocol.io) | [modelcontextprotocol/python-sdk](https://github.com/modelcontextprotocol/python-sdk) | Tier 1 |
13| [C#](https://csharp.sdk.modelcontextprotocol.io) | [modelcontextprotocol/csharp-sdk](https://github.com/modelcontextprotocol/csharp-sdk) | Tier 1 |
14| [Go](https://go.sdk.modelcontextprotocol.io) | [modelcontextprotocol/go-sdk](https://github.com/modelcontextprotocol/go-sdk) | Tier 1 |
15| [Java](https://java.sdk.modelcontextprotocol.io) | [modelcontextprotocol/java-sdk](https://github.com/modelcontextprotocol/java-sdk) | Tier 2 |
16| [Rust](https://rust.sdk.modelcontextprotocol.io) | [modelcontextprotocol/rust-sdk](https://github.com/modelcontextprotocol/rust-sdk) | Tier 2 |
17| Swift | [modelcontextprotocol/swift-sdk](https://github.com/modelcontextprotocol/swift-sdk) | Tier 3 |
18| [Ruby](https://ruby.sdk.modelcontextprotocol.io) | [modelcontextprotocol/ruby-sdk](https://github.com/modelcontextprotocol/ruby-sdk) | Tier 3 |
19| [PHP](https://php.sdk.modelcontextprotocol.io) | [modelcontextprotocol/php-sdk](https://github.com/modelcontextprotocol/php-sdk) | Tier 3 |
20| [Kotlin](https://kotlin.sdk.modelcontextprotocol.io) | [modelcontextprotocol/kotlin-sdk](https://github.com/modelcontextprotocol/kotlin-sdk) | Tier 3 |
2121
2222See [SDK Tiering System](/community/sdk-tiers) for details on what each tier means.
2323
from line 36
3636
3737Ready to start building with MCP? Choose your path:
3838
39<CardGroup cols={2}>
40 <Card title="Build a Server" icon="server" href="/docs/2024-11-05/develop/build-server">
41 Learn how to create your first MCP server
42 </Card>
39> **[Build a Server](/docs/2024-11-05/develop/build-server)**
40>
41> Learn how to create your first MCP server
4342
44 <Card title="Build a Client" icon="computer" href="/docs/2024-11-05/develop/build-client">
45 Create applications that connect to MCP servers
46 </Card>
47</CardGroup>
43> **[Build a Client](/docs/2024-11-05/develop/build-client)**
44>
45> Create applications that connect to MCP servers
4846
docs/2024-11-05/tools/debugging Changed · +46 / -54 lines
from line 32
3232logged to stderr (standard error) will be captured by the host application
3333automatically.
3434
35<Warning>
36 Local MCP servers should not log messages to stdout (standard out), as this
37 will interfere with protocol operation.
38</Warning>
35> [!WARNING]
36>
37> Local MCP servers should not log messages to stdout (standard out), as this
38> will interfere with protocol operation.
3939
4040For servers using the
4141[HTTP with SSE transport](/specification/2024-11-05/basic/transports#http-with-sse),
from line 46
4646For all [transports](/specification/2024-11-05/basic/transports), you can also
4747provide logging to the client by sending a log message notification:
4848
49<CodeGroup>
50 ```python Python theme={null}
51 @server.tool()
52 async def my_tool(ctx: Context) -> str:
53 await ctx.session.send_log_message(
54 level="info",
55 data="Server started successfully",
56 )
57 return "done"
58 ```
49```python Python theme={null}
50@server.tool()
51async def my_tool(ctx: Context) -> str:
52 await ctx.session.send_log_message(
53 level="info",
54 data="Server started successfully",
55 )
56 return "done"
57```
5958
60 ```typescript TypeScript theme={null}
61 await server.sendLoggingMessage({
62 level: "info",
63 data: "Server started successfully",
64 });
65 ```
66</CodeGroup>
59```typescript TypeScript theme={null}
60await server.sendLoggingMessage({
61 level: "info",
62 data: "Server started successfully",
63});
64```
6765
6866MCP defines eight
6967[RFC 5424 severity levels](/specification/2024-11-05/server/utilities/logging#log-levels)
from line 186
188186hover over the **Connectors** menu to see connected servers and available
189187tools.
190188
191<img src="https://mintcdn.com/mcp/zNouQwo2h8cbxlDS/images/available-mcp-tools.png?fit=max&auto=format&n=zNouQwo2h8cbxlDS&q=85&s=e2ace1ac88895a5fe30ebd8d01456bc3" alt="Available MCP tools" width="437" height="244" data-path="images/available-mcp-tools.png" />
189
192190
193191### Viewing logs
194192
from line 195
197195* macOS: `~/Library/Logs/Claude`
198196* Windows: `%APPDATA%\Claude\logs`
199197
200<CodeGroup>
201 ```bash macOS theme={null}
202 tail -n 20 -F ~/Library/Logs/Claude/mcp*.log
203 ```
198```bash macOS theme={null}
199tail -n 20 -F ~/Library/Logs/Claude/mcp*.log
200```
204201
205 ```powershell Windows theme={null}
206 type "$env:AppData\Claude\logs\mcp*.log"
207 ```
208</CodeGroup>
202```powershell Windows theme={null}
203type "$env:AppData\Claude\logs\mcp*.log"
204```
209205
210206The logs capture:
211207
from line 217
221217
2222181. Create a `developer_settings.json` file with `allowDevTools` set to true:
223219
224<CodeGroup>
225 ```bash macOS theme={null}
226 echo '{"allowDevTools": true}' > ~/Library/Application\ Support/Claude/developer_settings.json
227 ```
220```bash macOS theme={null}
221echo '{"allowDevTools": true}' > ~/Library/Application\ Support/Claude/developer_settings.json
222```
228223
229 ```powershell Windows theme={null}
230 '{"allowDevTools": true}' | Set-Content "$env:AppData\Claude\developer_settings.json"
231 ```
232</CodeGroup>
224```powershell Windows theme={null}
225'{"allowDevTools": true}' | Set-Content "$env:AppData\Claude\developer_settings.json"
226```
233227
2342282. Open DevTools: `Command-Option-I` (macOS) or `Ctrl+Alt+I` (Windows)
235229
from line 324
330324
331325## Next steps
332326
333<CardGroup cols={2}>
334 <Card title="MCP Inspector" icon="magnifying-glass" href="/docs/2024-11-05/tools/inspector">
335 Learn to use the MCP Inspector
336 </Card>
327> **[MCP Inspector](/docs/2024-11-05/tools/inspector)**
328>
329> Learn to use the MCP Inspector
337330
338 <Card title="Build an MCP server" icon="code" href="/docs/2024-11-05/develop/build-server">
339 Walk through building a server from scratch
340 </Card>
331> **[Build an MCP server](/docs/2024-11-05/develop/build-server)**
332>
333> Walk through building a server from scratch
341334
342 <Card title="Connect local servers" icon="plug" href="/docs/2024-11-05/develop/connect-local-servers">
343 Full claude\_desktop\_config.json reference and troubleshooting
344 </Card>
345</CardGroup>
335> **[Connect local servers](/docs/2024-11-05/develop/connect-local-servers)**
336>
337> Full claude\_desktop\_config.json reference and troubleshooting
346338