HomeDocs changes
Documentation changes
What the docs quietly changed.
Anthropic's published documentation, fetched on a schedule and diffed against the last copy. Every edit dated, including the ones that never made a release note.
Last poll read all 4 cleanly: Claude Code CLI, Claude Developer Platform, Claude Documentation, Model Context Protocol. A date here is when this site noticed, not when Anthropic edited.
Claude Developer Platform
808pages tracked
587 changed this week · last read 22 Oct 2026
Claude Code CLI
228pages tracked
146 changed this week · last read 22 Oct 2026
Claude Documentation
278pages tracked
74 changed this week · last read 22 Oct 2026
Model Context Protocol
357pages tracked
163 changed this week · last read 21 Oct 2026
100 on this page
9,144 of the ledger's changes in api. Show every change
ReadSourcePageWhatLines
21 Oct 2026
532 pages
+34650 −6634 lines
api
build-with-claude/prompt-engineering/prompting-claude-haiku-5-5
±changed
+4−4
apichanged
build-with-claude/prompt-engineering/prompting-claude-haiku-5-5
The model also sometimes needs an extra nudge to search. This happens most at `low` effort and with long system prompts. To fix it, add this text directly after the date:
The model also sometimes needs an extra nudge to search. This occurs most at `low` effort and with long system prompts. To fix it, add this text directly after the date:
Claude Haiku 5.5 sometimes writes reasoning-like text in the reply that users see. This happens more often with thinking off or at `low` effort. If you see this behavior, switch to
Claude Haiku 5.5 sometimes writes reasoning-like text in the reply that users see. This occurs more often with thinking off or at `low` effort. If you see this behavior, switch to
* `frontier_llm`: the request could assist the development of competing AI models.
+4 −4 lines
Recorded here 21 Oct 2026 19:37 UTC
api
build-with-claude/prompt-engineering/prompting-claude-fable-5-1
±changed
+18−18
apichanged
build-with-claude/prompt-engineering/prompting-claude-fable-5-1
* One tool call per turn in agent loops: [Batch independent tool calls in agent loops](https://platform.claude.com/docs/en/build-with-claude/prompt-engineering/prompting-claude-fab
* One tool call for each turn in agent loops: [Batch independent tool calls in agent loops](https://platform.claude.com/docs/en/build-with-claude/prompt-engineering/prompting-claud
Prompts, skills, and tool descriptions that ask the model to write out its thinking or reasoning may be declined with the `reasoning_extraction` category. Ask for a short explana
Prompts, skills, and tool descriptions that ask the model to write its thinking or reasoning might be declined with the `reasoning_extraction` category. Ask for a short explanati
First, check that your client receives progress updates at all. The model's short notes between tool calls, what it just found and what it's doing next, come back as [progress-upda
+18 −18 lines
Recorded here 21 Oct 2026 19:37 UTC
api
build-with-claude/prompt-engineering/prompting-claude-fable-5
±changed
+7−7
apichanged
build-with-claude/prompt-engineering/prompting-claude-fable-5
Claude Fable 5 has several behavioral differences from Claude Opus 4.8 that may require prompt or scaffolding updates. Capability improvements at this level are also a good prompt
Claude Fable 5 has several behavioral differences from Claude Opus 4.8 that might require prompt or scaffolding updates. Capability improvements at this level are also a good promp
Claude Fable 5 runs safety classifiers that target offensive cybersecurity techniques (such as building exploits, malware, or attack tooling), biology and life sciences content (
Claude Fable 5 runs safety classifiers that target offensive cybersecurity techniques (such as building exploits, malware, or attack tooling), biology and life sciences content (
* **Code review and debugging.** Bug-finding recall (outside the cybersecurity domains the safety classifiers cover) is noticeably higher than Claude Opus 4.8, including search acr
+7 −7 lines
Recorded here 21 Oct 2026 19:37 UTC
api
build-with-claude/prompt-engineering/overview
±changed
+2−2
apichanged
build-with-claude/prompt-engineering/overview
If not, spend time establishing that first. Check out [Define success criteria and build evaluations](https://platform.claude.com/docs/en/test-and-evaluate/develop-tests) for tips
If not, spend time establishing that first. See [Define success criteria and build evaluations](https://platform.claude.com/docs/en/test-and-evaluate/develop-tests) for tips and gu
This guide focuses on success criteria that are controllable through prompt engineering. Not every success criteria or failing eval is best solved by prompt engineering. For exampl
This guide focuses on success criteria that are controllable through prompt engineering. Not every success criterion or failing eval is best solved by prompt engineering. For examp
+2 −2 lines
Recorded here 21 Oct 2026 19:37 UTC
api
build-with-claude/prompt-engineering/claude-prompting-best-practices
±changed
+51−51
apichanged
build-with-claude/prompt-engineering/claude-prompting-best-practices
Claude responds well to clear, explicit instructions. Being specific about your desired output can help enhance results. If you want "above and beyond" behavior, explicitly request
Claude responds well to clear, explicit instructions. Being specific about the output you want can help enhance results. If you want Claude to do more than the task strictly requir
**Golden rule:** Show your prompt to a colleague with minimal context on the task and ask them to follow it. If they'd be confused, Claude will be too.
* Be specific about the desired output format and constraints.
+51 −51 lines
Recorded here 21 Oct 2026 19:37 UTC
api
build-with-claude/prompt-caching
±changed
+65−65
apichanged
build-with-claude/prompt-caching
* **[Automatic caching](https://platform.claude.com/docs/en/build-with-claude/prompt-caching#automatic-caching)**: Add a single `cache_control` field at the top level of your reque
* **[Explicit cache breakpoints](https://platform.claude.com/docs/en/build-with-claude/prompt-caching#explicit-cache-breakpoints)**: Place `cache_control` directly on individual co
* **[Automatic caching](https://platform.claude.com/docs/en/build-with-claude/prompt-caching#automatic-caching):** Add a single `cache_control` field at the top level of your reque
* **[Explicit cache breakpoints](https://platform.claude.com/docs/en/build-with-claude/prompt-caching#explicit-cache-breakpoints):** Place `cache_control` directly on individual co
3. Otherwise, it processes the full prompt and caches the prefix once the response begins.
+65 −65 lines
Recorded here 21 Oct 2026 19:37 UTC
api
build-with-claude/preserved-thinking
±changed
+43−44
apichanged
build-with-claude/preserved-thinking
* [Drops some `thinking` blocks and keeps later ones](https://platform.claude.com/docs/en/build-with-claude/preserved-thinking#append-assistant-turns-exactly-as-returned), or remov
* [Drops some `thinking` blocks and keeps later ones](https://platform.claude.com/docs/en/build-with-claude/preserved-thinking#append-assistant-turns-exactly-as-returned), or remov
Keep sending the full history on every request, thinking blocks included, and let the API drop what the current model can't read. The API never edits your `messages` array, so the
Keep sending the full history on every request, thinking blocks included, and let the API drop what the current model can't read. The API never edits your `messages` array, so the
Keep `system` and `tools` fixed for the session and treat `messages` as append-only. The same discipline keeps the prefix stable for [prompt caching](https://platform.claude.com/do
+43 −44 lines
Recorded here 21 Oct 2026 19:37 UTC
api
build-with-claude/pdf-support
±changed
+22−22
apichanged
build-with-claude/pdf-support
* Assisting with document translation
* Helping with document translation
## Before you begin
## Before you start
| Requirement | Limit |
+22 −22 lines
Recorded here 21 Oct 2026 19:37 UTC
api
build-with-claude/overview
±changed
+18−18
apichanged
build-with-claude/overview
* **Tool infrastructure:** Handles discovery and orchestration at scale.
* **Context management:** Keeps long-running sessions efficient.
* **Tool infrastructure:** Handle discovery and orchestration at scale.
* **Context management:** Keep long-running sessions efficient.
| [Adaptive thinking](https://platform.claude.com/docs/en/build-with-claude/thinking) | Let Claude dynamically determine when and how much to think. The only thinki
+18 −18 lines
Recorded here 21 Oct 2026 19:37 UTC
api
build-with-claude/multilingual-support
±changed
+1−1
apichanged
build-with-claude/multilingual-support
1. **Provide clear language context:** Although Claude can detect the target language automatically, explicitly stating the desired input and output languages improves reliability.
1. **Provide clear language context:** Although Claude can detect the target language automatically, explicitly stating the input and output languages you want improves reliability
+1 −1 lines
Recorded here 21 Oct 2026 19:37 UTC
api
build-with-claude/mid-conversation-system-messages
±changed
+15−15
apichanged
build-with-claude/mid-conversation-system-messages
* **Declare what you know up front.** A tool you know about at the first request belongs in `tools`, with `defer_loading: true` and a later `tool_addition` reference if the model s
* **Declare what you know up front.** A tool you know about at the first request belongs in `tools`, with `defer_loading: true` and a later `tool_addition` reference if it shouldn'
* The tool definitions sent after the first user message that are still available after any message total more than 4 MB (4,194,304 bytes).
* The rendered tool text is larger than 4 MB (4,194,304 bytes).
* The tool definitions sent after the first user message that are still available after any message total more than 4 MB (4,194,304 bytes).
+15 −15 lines
Recorded here 21 Oct 2026 19:37 UTC
api
build-with-claude/mid-conversation-effort-example
±changed
+5−5
apichanged
build-with-claude/mid-conversation-effort-example
The example is a single file. The constants control the effort level, the fan-out shape, and how often the mode refresher is re-sent. `MAX_CONCURRENT` caps how many subagents run a
The example is a single file. The constants control the effort level, the fan-out shape, and how often the mode refresher is re-sent. `MAX_CONCURRENT` caps how many subagents run a
A fan-out that spawns dozens of subagents is expensive to restart from scratch. A small content-addressed journal makes it idempotent: before dispatching a subagent, look up the SH
A fan-out that spawns dozens of subagents is expensive to restart from the beginning. A small content-addressed journal makes it idempotent: before dispatching a subagent, look up
The fan-out accepts up to `MAX_TOTAL_SUBTASKS` prompts, runs them through the journal with at most `MAX_CONCURRENT` in flight (sequential in the PHP port), and isolates failures so
+5 −5 lines
Recorded here 21 Oct 2026 19:37 UTC
api
build-with-claude/handling-stop-reasons
±changed
+18−18
apichanged
build-with-claude/handling-stop-reasons
| [`end_turn`](https://platform.claude.com/docs/en/build-with-claude/handling-stop-reasons#end-turn) | Claude finished its response natura
| [`end_turn`](https://platform.claude.com/docs/en/build-with-claude/handling-stop-reasons#end-turn) | Claude completed its response natur
The most common stop reason. Indicates Claude finished its response naturally.
The most common stop reason. Indicates Claude completed its response naturally.
Sometimes Claude returns an empty response (exactly 2–3 tokens with no content) with `stop_reason: "end_turn"`. This typically occurs when Claude interprets that the assistant tu
+18 −18 lines
Recorded here 21 Oct 2026 19:37 UTC
api
build-with-claude/files
±changed
+33−37
apichanged
build-with-claude/files
**Uploaded files are accessible to your entire workspace, not scoped to an end user, conversation, or session.** Any API key with access to a workspace can access any files uploa
If you are building a multi-tenant application on the Files API, create a separate [workspace](https://platform.claude.com/docs/en/manage-claude/workspaces) for each tenant. The
**Uploaded files are accessible to your entire workspace, not scoped to an end user, conversation, or session.** Any API key with access to a workspace can access any files uploa
+33 −37 lines
Recorded here 21 Oct 2026 19:37 UTC
api
build-with-claude/fast-mode
±changed
+3−3
apichanged
build-with-claude/fast-mode
model: "claude-opus-5-5",
model: Anthropic::Model::CLAUDE_OPUS_5_5,
model: "claude-opus-5-5",
model: Anthropic::Model::CLAUDE_OPUS_5_5,
Falling back from fast to standard speed will result in a [prompt cache](https://platform.claude.com/docs/en/build-with-claude/prompt-caching) miss. Requests at different speeds
+3 −3 lines
Recorded here 21 Oct 2026 19:37 UTC
api
build-with-claude/fallback-credit
±changed
+18−18
apichanged
build-with-claude/fallback-credit
Prompt caches are per-model. When a model declines a request and you retry on another model, the conversation prefix already cached for the first model must be written into the new
Prompt caches are per-model. When a model declines a request and you retry on another model, the conversation prefix already cached for the first model must be written into the new
* **`fallback_credit_token`:** an opaque string that represents the credit.
* **`fallback_has_prefill_claim`:** a Boolean that tells you which retry body shape to use.
* **`fallback_credit_token`:** an opaque string that represents the credit
+18 −18 lines
Recorded here 21 Oct 2026 19:37 UTC
api
build-with-claude/extended-thinking
±changed
+9−9
apichanged
build-with-claude/extended-thinking
Extended thinking availability per model, including the models where extended thinking is the only mode, is listed in the [per-model configuration table](https://platform.claude.co
Extended thinking availability for each model, including the models where extended thinking is the only mode, is listed in the [per-model configuration table](https://platform.clau
The `budget_tokens` parameter sets a target for how many tokens Claude can use for its internal reasoning process. Larger budgets can improve response quality by enabling more thor
The `budget_tokens` parameter sets a target for how many tokens Claude can use for its internal reasoning process. Larger budgets can improve response quality by allowing more thor
* **Less than `max_tokens`.** Thinking tokens count toward the `max_tokens` limit for the turn, so the budget must leave room for the final response. The one exception is [interlea
+9 −9 lines
Recorded here 21 Oct 2026 19:37 UTC
api
build-with-claude/embeddings
±changed
+8−8
apichanged
build-with-claude/embeddings
description: Text embeddings are numerical representations of text that enable measuring semantic similarity. This guide introduces embeddings, their applications, and how to use e
description: Text embeddings are numerical representations of text that let you measure semantic similarity. This guide introduces embeddings, their applications, and how to use em
* Dataset size & domain specificity: size of the model training dataset and its relevance to the domain you want to embed. Larger or more domain-specific data generally produces be
* Inference performance: embedding lookup speed and end-to-end latency. This is a particularly important consideration for large scale production deployments
* Dataset size and domain specificity: size of the model training dataset and its relevance to the domain you want to embed. Larger or more domain-specific data generally produces
+8 −8 lines
Recorded here 21 Oct 2026 19:37 UTC
api
build-with-claude/effort
±changed
+9−9
apichanged
build-with-claude/effort
model: "claude-opus-5-5",
model: Anthropic::Model::CLAUDE_OPUS_5_5,
Because effort applies to every output token, it works whether or not thinking is enabled. Lower effort also means fewer and terser tool calls.
Because effort applies to every output token, it works with or without thinking. Lower effort also means fewer and terser tool calls.
Claude Opus 4.7 also respects effort levels more strictly than Claude Opus 4.6, especially at `low` and `medium`. At lower effort levels, the model scopes its work to what was aske
+9 −9 lines
Recorded here 21 Oct 2026 19:37 UTC
api
build-with-claude/context-windows
±changed
+2−2
apichanged
build-with-claude/context-windows
+262 −260 lines
Recorded here 21 Oct 2026 19:37 UTC
api
build-with-claude/compaction-threshold
±changed
+17−17
apichanged
build-with-claude/compaction-threshold
model: "claude-opus-5-5",
model: Anthropic::Model::CLAUDE_OPUS_5_5,
model: "claude-opus-5-5",
model: Anthropic::Model::CLAUDE_OPUS_5_5,
model: "claude-opus-5-5",
+17 −17 lines
Recorded here 21 Oct 2026 19:37 UTC
api
build-with-claude/compaction-thinking-blocks
±changed
+2−2
apichanged
build-with-claude/compaction-thinking-blocks
A later request can use a different `system`, different `tools`, or a different model than the compaction request, and the API still accepts the block. Such a change can invalidate
A later request can use a different `system`, different `tools`, or a different model from the compaction request, and the API still accepts the block. Such a change can invalidate
The `prefix_mismatch_behavior` field needs the `thinking-binding-controls-2026-08-01` beta header in addition to [the `compact-2026-09-04` beta header](https://platform.claude.com/
The `prefix_mismatch_behavior` field needs the `thinking-binding-controls-2026-08-01` beta header in addition to the [`compact-2026-09-04` beta header](https://platform.claude.com/
+2 −2 lines
Recorded here 21 Oct 2026 19:37 UTC
api
build-with-claude/compaction-on-demand
±changed
+17−17
apichanged
build-with-claude/compaction-on-demand
With on-demand compaction, your application decides when a conversation is summarized: you send one request with the `compaction` parameter, and Claude returns a summary in place o
With on-demand compaction, your application determines when a conversation is summarized: you send one request with the `compaction` parameter, and Claude returns a summary in plac
From then on the block takes the place of the messages it summarizes. It goes first in `messages`, the summarized messages are removed, and your next turn follows it. Claude sees t
From then on the block takes the place of the messages it summarizes. It goes first in `messages`, the summarized messages are removed, and your next turn follows it. Claude receiv
. For wo
Anthropic uses AWS infrastructure to process your API requests, run model inference, and store your workspace content (such as prompts, outputs, files, Skills, and batches). For wo
* **Claude Platform on AWS** (this page): The Claude API platform billed through AWS Marketplace. Managed in the [Claude Console](https://platform.claude.com/docs/en/build-with-cla
* **Claude Platform on AWS (this page):** The Claude API platform billed through AWS Marketplace. Managed in the [Claude Console](https://platform.claude.com/docs/en/build-with-cla
Setting up Claude Platform on AWS happens in four phases: sign up on the AWS Console service page, complete your Anthropic organization setup, note your workspace ID, and sign in t
+52 −52 lines
Recorded here 21 Oct 2026 19:37 UTC
api
build-with-claude/claude-on-vertex-ai
±changed
+7−7
apichanged
build-with-claude/claude-on-vertex-ai
Agent Platform limits request payloads to 30 MB. When sending large documents or many images, you might reach this limit before the token limit.
Agent Platform limits request payloads to 30 MB. When sending large documents or many images, you might reach this limit before the token limit.
Regional and multi-region endpoints include a 10% pricing premium over global endpoints.
Regional and multi-region endpoints include a 10 percent pricing premium over global endpoints.
* No pricing premium
+7 −7 lines
Recorded here 21 Oct 2026 19:37 UTC
api
build-with-claude/claude-on-amazon-bedrock-legacy
±changed
+14−14
apichanged
build-with-claude/claude-on-amazon-bedrock-legacy
2. Configure your AWS credentials using the AWS configure command (see [Configure the AWS CLI](https://docs.aws.amazon.com/cli/latest/userguide/cli-chap-configure.html)) or find yo
2. Configure your AWS credentials using the AWS configure command (see [Configure the AWS CLI](https://docs.aws.amazon.com/cli/latest/userguide/cli-chap-configure.html)) or find yo
Anthropic's [client SDKs](https://platform.claude.com/docs/en/cli-sdks-libraries/overview) support Bedrock. You can also use an AWS SDK like `boto3` directly.
Anthropic's [client SDKs](https://platform.claude.com/docs/en/cli-sdks-libraries/overview) support Bedrock. You can also use an AWS SDK such as `boto3` directly.
<CodeGroup>
+14 −14 lines
Recorded here 21 Oct 2026 19:37 UTC
api
build-with-claude/claude-in-microsoft-foundry
±changed
+7−7
apichanged
build-with-claude/claude-in-microsoft-foundry
Before you begin, ensure you have:
Before you start, ensure you have:
Anthropic's [client SDKs](https://platform.claude.com/docs/en/cli-sdks-libraries/overview) support Foundry through a platform-specific package or client class. The examples on this
Anthropic's [client SDKs](https://platform.claude.com/docs/en/cli-sdks-libraries/overview) support Foundry through a platform-specific package or client class. The examples on this
Foundry uses a two-level hierarchy: **resources** contain your security and billing configuration, while **deployments** are the model instances you call through the API. You'll fi
+7 −7 lines
Recorded here 21 Oct 2026 19:37 UTC
api
build-with-claude/claude-in-amazon-bedrock
±changed
+6−6
apichanged
build-with-claude/claude-in-amazon-bedrock
Before you begin, ensure you have:
* An AWS account with [Amazon Bedrock model access](https://console.aws.amazon.com/bedrock/home#/modelaccess) enabled for the Claude models you intend to use.
* The [AWS CLI](https://docs.aws.amazon.com/cli/latest/userguide/getting-started-install.html) installed and configured (optional, for credential management).
Before you start, ensure you have:
+6 −6 lines
Recorded here 21 Oct 2026 19:37 UTC
api
build-with-claude/citations
±changed
+3−3
apichanged
build-with-claude/citations
* **Cost savings:** If your prompt-based approach asks Claude to output direct quotes, you may see cost savings because `cited_text` does not count toward your output tokens.
* **Cost savings:** If your prompt-based approach asks Claude to output direct quotes, you might see cost savings because `cited_text` does not count toward your output tokens.
* Responses may now include multiple text blocks where each text block can contain a claim that Claude is making and a list of citations that support the claim.
* Responses might now include multiple text blocks where each text block can contain a claim that Claude is making and a list of citations that support the claim.
PDF documents can be provided as base64-encoded data, a URL, or by `file_id`. PDF text is extracted and chunked into sentences. As image citations are not yet supported, PDFs that
+3 −3 lines
Recorded here 21 Oct 2026 19:37 UTC
api
build-with-claude/cache-diagnostics
±changed
+7−7
apichanged
build-with-claude/cache-diagnostics
This workflow doesn't translate well to a one-off shell command. See the SDK tabs for the loop pattern; the per-turn HTTP request is identical to [Basic usage](https://platfo
This workflow doesn't translate well to a one-off shell command. Use the Claude SDK to run the loop; the per-turn HTTP request is identical to [Basic usage](https://platform.
This workflow doesn't translate well to a one-off shell command. See the SDK tabs for the loop pattern; the per-turn CLI invocation is identical to [Basic usage](https://plat
This workflow doesn't translate well to a one-off shell command. Use the Claude SDK to run the loop; the per-turn CLI command is identical to [Basic usage](https://platform.c
| `{"cache_miss_reason": null}` | The comparison was still running when the response was serialized. This can happen when the response starts very quickly. Treat it as inconclusiv
+7 −7 lines
Recorded here 21 Oct 2026 19:37 UTC
api
build-with-claude/batch-processing
±changed
+121−123
apichanged
build-with-claude/batch-processing
description: Process large volumes of Messages requests asynchronously with the Message Batches API, cutting costs by 50% and increasing throughput.
description: Process large volumes of Messages requests asynchronously with the Message Batches API, cutting costs by 50 percent and increasing throughput.
* Batch results are available for 29 days after creation. After that, you may still view the Batch, but its results will no longer be available for download.
* Batch results are available for 29 days after creation. After that, you can still view the Batch, but its results will no longer be available for download.
* Rate limits apply to both Batches API HTTP requests and the number of requests within a batch waiting to be processed. See [Message Batches API rate limits](https://platform.clau
+121 −123 lines
Recorded here 21 Oct 2026 19:37 UTC
api
api/supported-regions
±changed
+1−1
apichanged
api/supported-regions
description: "Here are the countries, regions, and territories we can currently support access from:"
description: "Here are the countries, regions, and territories Anthropic can support access from:"
+1 −1 lines
Recorded here 21 Oct 2026 19:37 UTC
api
api/service-tiers
±changed
+4−4
apichanged
api/service-tiers
* Cache writes as 1.25 tokens per token written to the cache with a 5 minute TTL
* Cache writes as 2.00 tokens per token written to the cache with a 1 hour TTL
* Cache writes as 1.25 tokens per token written to the cache with a 5-minute TTL
* Cache writes as 2.00 tokens per token written to the cache with a 1-hour TTL
* `"standard_only"` - Only use standard tier capacity, useful if you don't want to use your Priority Tier capacity
+4 −4 lines
Recorded here 21 Oct 2026 19:37 UTC
api
api/rate-limits
±changed
+16−16
apichanged
api/rate-limits
The API enforces service-configured limits at the organization level, but you may also set user-configurable limits for your organization's workspaces.
The API enforces service-configured limits at the organization level, but you can also set user-configurable limits for your organization's workspaces.
When usage reaches a spend limit you set, requests return HTTP 400 with error type `invalid_request_error`. The message begins `You have reached your specified API usage limits`, o
When usage reaches a spend limit you set, requests return HTTP 400 with error type `invalid_request_error`. The message starts with `You have reached your specified API usage limit
The rate limits for the Messages API are measured in requests per minute (RPM), input tokens per minute (ITPM), and output tokens per minute (OTPM) for each model class. If you exc
+16 −16 lines
Recorded here 21 Oct 2026 19:37 UTC
api
api/overview
±changed
+29−29
apichanged
api/overview
* **[Messages API](https://platform.claude.com/docs/en/api/messages/create)**: Send messages to Claude for conversational interactions (`POST /v1/messages`)
* **[Message Batches API](https://platform.claude.com/docs/en/api/messages/batches/create)**: Process large volumes of Messages requests asynchronously with 50% cost reduction (`PO
* **[Token Counting API](https://platform.claude.com/docs/en/api/messages/count_tokens)**: Count tokens in a message before sending to manage costs and rate limits (`POST /v1/messa
* **[Models API](https://platform.claude.com/docs/en/api/models/list)**: List available Claude models and their details (`GET /v1/models`)
* **[Files API](https://platform.claude.com/docs/en/api/files/upload)**: Upload and manage files for use across multiple API calls (`POST /v1/files`, `GET /v1/files`)
+29 −29 lines
Recorded here 21 Oct 2026 19:37 UTC
api
api/opentelemetry/spans
±changed
+14−14
apichanged
api/opentelemetry/spans
Attributes tagged with <Badge>Content</Badge> are only on spans in content mode, from the SDK and the API alike. See [how the content modes differ in the guide](https://platform.cl
Attributes tagged with <Badge>Content</Badge> are only on spans in content mode, from the SDK and the API alike. See [how the content modes differ in the guide](https://platform.cl
* The [`anthropic.sessions.update`](https://platform.claude.com/docs/en/api/opentelemetry/spans#anthropic-sessions-update) span, for a turn that resumes because the session budge
* The [`anthropic.sessions.update`](https://platform.claude.com/docs/en/api/opentelemetry/spans#anthropic-sessions-update) span, for a turn that resumes because the request raise
| Attribute | Type | Description
+14 −14 lines
Recorded here 21 Oct 2026 19:37 UTC
api
api/ip-addresses
±changed
+2−2
apichanged
api/ip-addresses
description: Anthropic services use fixed IP addresses for both inbound and outbound connections. You can use these addresses to configure your firewall rules for secure access to
description: Anthropic services use fixed IP addresses for both inbound and outbound connections. You can use these addresses to configure your firewall rules for secure access to
The following IP addresses are no longer in use by Anthropic. If you have previously allowlisted these addresses, you should remove them from your firewall rules.
The following IP addresses are no longer in use by Anthropic. If you have previously allowlisted these addresses, remove them from your firewall rules.
+2 −2 lines
Recorded here 21 Oct 2026 19:37 UTC
api
api/errors
±changed
+15−15
apichanged
api/errors
529 errors can occur when the API experiences high traffic across all users.
In rare cases, if your organization has a sharp increase in usage, you might see 429 errors because of acceleration limits on the API. To avoid hitting acceleration limits, ram
A 529 error can occur when the API experiences high traffic across all users.
+15 −15 lines
Recorded here 21 Oct 2026 19:37 UTC
api
api/claude-code/routines-fire
±changed
+2−2
apichanged
api/claude-code/routines-fire
## Before you begin
## Before you start
* [Errors](https://platform.claude.com/docs/en/api/errors)
* [Claude API errors](https://platform.claude.com/docs/en/api/errors)
+2 −2 lines
Recorded here 21 Oct 2026 19:37 UTC
api
api/beta-headers
±changed
+6−6
apichanged
api/beta-headers
Each feature's documentation states the exact beta name to send. The [API overview](https://platform.claude.com/docs/en/api/overview) lists the APIs currently in beta.
The following examples show the same request with cURL, the `ant` CLI, and the SDK, using the [context editing](https://platform.claude.com/docs/en/build-with-claude/context-editin
Each feature's documentation states the exact beta name to send. The [API overview](https://platform.claude.com/docs/en/api/overview) lists the APIs in beta.
+6 −6 lines
Recorded here 21 Oct 2026 19:37 UTC
api
agents-and-tools/tool-use/web-search-tool
±changed
+10−10
apichanged
agents-and-tools/tool-use/web-search-tool
model: "claude-opus-5-5",
model: Anthropic::Model::CLAUDE_OPUS_5_5,
model: "claude-opus-5-5",
model: Anthropic::Model::CLAUDE_OPUS_5_5,
All web search tool versions accept `allowed_callers`, which controls whether Claude calls web search directly or from code execution through [dynamic filtering](https://platform.c
+10 −10 lines
Recorded here 21 Oct 2026 19:37 UTC
api
agents-and-tools/tool-use/web-fetch-tool
±changed
+11−11
apichanged
agents-and-tools/tool-use/web-fetch-tool
The web fetch tool allows Claude to retrieve full content from specified web pages and PDF documents.
The web fetch tool allows Claude to retrieve full content from specified webpages and PDF documents.
Fetching full web pages and PDFs can quickly consume tokens, especially when only specific information is needed from large documents. With `web_fetch_20260209` or later, Claude ca
Fetching full webpages and PDFs can quickly consume tokens, especially when only specific information is needed from large documents. With `web_fetch_20260209` or later, Claude can
* Processing structured data from web pages
+11 −11 lines
Recorded here 21 Oct 2026 19:37 UTC
api
agents-and-tools/tool-use/troubleshooting-tool-use
±changed
+5−5
apichanged
agents-and-tools/tool-use/troubleshooting-tool-use
| Claude never calls your tool | Tool name collision or overly-generic schema | Check for duplicate names across your tool list. Add `input_examples` to make the inte
| Claude never calls your tool | Tool name collision or overly generic schema | Check for duplicate names across your tool list. Add `input_examples` to make the inte
| `disable_parallel_tool_use` seems ignored | Set too late in the conversation | Must be set on the request that returns `tool_use`. Setting it on a later reque
| `disable_parallel_tool_use` seems ignored | Set too late in the conversation | Set it on the request that returns `tool_use`. Setting it on a later request ha
| `tool_use ids were found without tool_result blocks immediately after` | Missing `tool_result` for some `tool_use` ids, or `tool_result` is not the first content block in the use
+5 −5 lines
Recorded here 21 Oct 2026 19:37 UTC
api
agents-and-tools/tool-use/tool-use-with-prompt-caching
±changed
+2−2
apichanged
agents-and-tools/tool-use/tool-use-with-prompt-caching
The [computer use](https://platform.claude.com/docs/en/agents-and-tools/tool-use/computer-use-tool) and [browser use](https://platform.claude.com/docs/en/agents-and-tools/tool-use/
The [computer use](https://platform.claude.com/docs/en/agents-and-tools/tool-use/computer-use-tool) and [browser use](https://platform.claude.com/docs/en/agents-and-tools/tool-use/
This automatic breakpoint always uses the default 5-minute TTL, independent of any TTL you set on your own `cache_control` markers. In the response `usage`, these writes appear und
This automatic breakpoint always uses the default 5-minute TTL, independent of any TTL you set on your own `cache_control` markers. In the response `usage`, these writes appear und
+2 −2 lines
Recorded here 21 Oct 2026 19:37 UTC
api
agents-and-tools/tool-use/tool-search-tool
±changed
+6−6
apichanged
agents-and-tools/tool-use/tool-search-tool
5. The API runs the search and returns the matching tools as `tool_reference` blocks (up to 5 by default; Claude can set a `limit` in its search input).
5. The API runs the search and returns the matching tools as `tool_reference` blocks (up to 5 by default, although Claude can set a `limit` in its search input).
Claude searches the catalog, discovers `get_weather`, and calls it. The response ends with `stop_reason: "tool_use"`. Execute the discovered tool and return a `tool_result` as in [
Claude searches the catalog, discovers `get_weather`, and calls it. The response ends with `stop_reason: "tool_use"`. Run the discovered tool and return a `tool_result` as in [Hand
The computer use and browser use toolsets (`computer_toolset_20260801` and `browser_toolset_20260801`) take `defer_loading` per member tool inside the entry's `configs` object, not
+6 −6 lines
Recorded here 21 Oct 2026 19:37 UTC
api
agents-and-tools/tool-use/tool-runner
±changed
+6−6
apichanged
agents-and-tools/tool-use/tool-runner
**Using Zod (recommended)** - Use `betaZodTool()` for type-safe tool definitions with Zod validation (requires Zod 3.25.0 or higher):
**Using Zod (recommended):** Use `betaZodTool()` for type-safe tool definitions with Zod validation (requires Zod 3.25.0 or higher):
**Using JSON Schema** - Use `betaTool()` for type-safe tool definitions without Zod:
**Using JSON Schema:** Use `betaTool()` for type-safe tool definitions without Zod:
When you take over for an iteration, the runner does not append the assistant message or tool results from that turn. You become responsible for keeping the conversation valid: app
+6 −6 lines
Recorded here 21 Oct 2026 19:37 UTC
api
agents-and-tools/tool-use/tool-reference
±changed
+6−6
apichanged
agents-and-tools/tool-use/tool-reference
Anthropic provides two kinds of tools: **server tools** that execute on Anthropic's infrastructure, and **client tools** where Anthropic defines the schema but your application han
Anthropic provides two kinds of tools: **server tools** that run on Anthropic's infrastructure, and **client tools** where Anthropic defines the schema but your application handles
The [computer use tool](https://platform.claude.com/docs/en/agents-and-tools/tool-use/computer-use-tool) and [browser use tool](https://platform.claude.com/docs/en/agents-and-tools
The [computer use tool](https://platform.claude.com/docs/en/agents-and-tools/tool-use/computer-use-tool) and [browser use tool](https://platform.claude.com/docs/en/agents-and-tools
* A disabled member is removed from the tools Claude sees. If Claude still names it, return an error `tool_result`.
+6 −6 lines
Recorded here 21 Oct 2026 19:37 UTC
api
agents-and-tools/tool-use/tool-combinations
±changed
+4−4
apichanged
agents-and-tools/tool-use/tool-combinations
Anthropic-provided tools are designed to work together. Common agent patterns pair tools that cover complementary stages of a workflow: one tool gathers or discovers, another proce
Anthropic-provided tools are designed to work together. Common agent patterns pair tools that cover complementary stages of a workflow: one tool gathers or discovers, another proce
The flow is typically search, then execute, then optionally search again if the first pass surfaced a gap. Code execution runs server-side, so there's no client-side sandbox to man
The flow is typically search, then run code, then optionally search again if the first pass surfaced a gap. Code execution runs server-side, so there's no client-side sandbox to ma
The computer use tool subsumes most others by operating a full desktop. Claude sees screenshots and issues mouse and keyboard actions, which means it can drive any application a hu
+4 −4 lines
Recorded here 21 Oct 2026 19:37 UTC
api
agents-and-tools/tool-use/text-editor-tool
±changed
+29−29
apichanged
agents-and-tools/tool-use/text-editor-tool
* Include the text editor tool in your API request
* Provide a user prompt that may require examining or modifying files, such as "Can you fix the syntax error in my code?"
* Include the text editor tool in your API request.
* Provide a user prompt that might require examining or modifying files, such as "Can you fix the syntax error in my code?"
* Claude assesses what it needs to look at and uses the `view` command to examine file contents or list directory contents
+29 −29 lines
Recorded here 21 Oct 2026 19:37 UTC
api
agents-and-tools/tool-use/strict-tool-use
±changed
+2−2
apichanged
agents-and-tools/tool-use/strict-tool-use
Strict tool use validates tool parameters, ensuring Claude calls your functions with correctly-typed arguments. Use strict tool use when you need to:
Strict tool use validates tool parameters, ensuring Claude calls your functions with correctly typed arguments. Use strict tool use when you need to:
* Functions receive correctly-typed arguments every time
* Functions receive correctly typed arguments every time
+2 −2 lines
Recorded here 21 Oct 2026 19:37 UTC
api
agents-and-tools/tool-use/server-tools
±changed
+11−11
apichanged
agents-and-tools/tool-use/server-tools
description: "Work with Anthropic-executed tools: server_tool_use blocks, pause_turn continuation, mixed server and client tool turns, and domain filtering."
description: "Work with tools that Anthropic runs: server_tool_use blocks, pause_turn continuation, mixed server and client tool turns, and domain filtering."
Server-executed tools share these mechanics: the `server_tool_use` block, `pause_turn` continuation, turns that mix server and client tools, Zero Data Retention (ZDR) eligibility,
Server-executed tools share these mechanics: the `server_tool_use` block, `pause_turn` continuation, turns that mix server and client tools, Zero Data Retention (ZDR) eligibility,
The API executes the tool internally. You see the call and its result in the response, but you don't handle execution. Unlike client `tool_use` blocks, you don't need to respond wi
+11 −11 lines
Recorded here 21 Oct 2026 19:37 UTC
api
agents-and-tools/tool-use/programmatic-tool-calling
±changed
+80−80
apichanged
agents-and-tools/tool-use/programmatic-tool-calling
Programmatic tool calling allows Claude to write code that calls your tools programmatically within a [code execution](https://platform.claude.com/docs/en/agents-and-tools/tool-use
Consider checking budget compliance across 20 employees: the traditional approach requires 20 separate model round-trips, pulling thousands of expense line items into the context a
Programmatic tool calling allows Claude to write code that calls your tools programmatically within a [code execution](https://platform.claude.com/docs/en/agents-and-tools/tool-use
+80 −80 lines
Recorded here 21 Oct 2026 19:37 UTC
api
agents-and-tools/tool-use/parallel-tool-use
±changed
+8−8
apichanged
agents-and-tools/tool-use/parallel-tool-use
model: "claude-opus-5-5",
model: Anthropic::Model::CLAUDE_OPUS_5_5,
model: "claude-opus-5-5",
model: Anthropic::Model::CLAUDE_OPUS_5_5,
When `tool_choice` type is `auto` (the default), setting `disable_parallel_tool_use: true` means Claude calls at most one tool per response. Claude can still answer in plain text w
+8 −8 lines
Recorded here 21 Oct 2026 19:37 UTC
api
agents-and-tools/tool-use/overview
±changed
+12−12
apichanged
agents-and-tools/tool-use/overview
description: Connect Claude to external tools and APIs. See where tools execute, when Claude calls them, and which tool fits your task.
description: Connect Claude to external tools and APIs. See where tools run, when Claude calls them, and which tool fits your task.
Tool use (also called function calling) lets Claude call functions that you define or that Anthropic provides. Claude determines when to call a tool based on the user's request and
Here's a minimal example using a server tool, the [Web search tool](https://platform.claude.com/docs/en/agents-and-tools/tool-use/web-search-tool), which Anthropic executes for you
+12 −12 lines
Recorded here 21 Oct 2026 19:37 UTC
api
agents-and-tools/tool-use/memory-tool
±changed
+10−10
apichanged
agents-and-tools/tool-use/memory-tool
The memory tool operates client-side: Claude requests file operations, and your application executes them. You control where and how the data is stored through your own infrastruct
The memory tool operates client-side: Claude requests file operations, and your application runs them. You control where and how the data is stored through your own infrastructure.
Reach out through the [feedback form](https://forms.gle/YXC2EKGMhjN1c4L88) to share your feedback on this feature.
Use the [feedback form](https://forms.gle/YXC2EKGMhjN1c4L88) to share your feedback on this feature.
Because the memory tool is client-side, Claude only requests memory operations. Your application executes each request against storage you control and returns the result in a `tool
+10 −10 lines
Recorded here 21 Oct 2026 19:37 UTC
api
agents-and-tools/tool-use/manage-tool-context
±changed
+5−5
apichanged
agents-and-tools/tool-use/manage-tool-context
description: Choose between tool search, programmatic tool calling, prompt caching, and context editing to manage context bloat.
description: Choose among tool search, programmatic tool calling, prompt caching, and context editing to manage context bloat.
Tool definitions and accumulated `tool_result` blocks consume your context window. Long-running agents with many tools or many turns can exhaust available context before the task i
Tool definitions and accumulated `tool_result` blocks consume your context window. Long-running agents with many tools or many turns can exhaust available context before the task i
| Programmatic tool calling | `tool_result` roundtrips | Chains of tool calls that can execute as a single script | [Programmatic tool calling](https://p
+5 −5 lines
Recorded here 21 Oct 2026 19:37 UTC
api
agents-and-tools/tool-use/how-tool-use-works
±changed
+10−10
apichanged
agents-and-tools/tool-use/how-tool-use-works
description: Understand the tool use loop, where tools execute, and when to use tools instead of prose.
description: Understand the tool use loop, where tools run, and when to use tools instead of prose.
Tool use is a contract between your application and the model. You specify what operations are available and what shape their inputs and outputs take; Claude determines when and ho
Tool use is a contract between your application and the model. You specify what operations are available and what shape their inputs and outputs take; Claude determines when and ho
The primary axis along which tools differ is where the code executes. Every tool falls into one of three buckets, and the bucket determines what your application is responsible for
+10 −10 lines
Recorded here 21 Oct 2026 19:37 UTC
api
agents-and-tools/tool-use/handle-tool-calls
±changed
+19−19
apichanged
agents-and-tools/tool-use/handle-tool-calls
The response will have a `stop_reason` of `tool_use` and one or more `tool_use` content blocks that include:
* `id`: A unique identifier for this particular tool use block. This will be used to match up the tool results later.
The response has a `stop_reason` of `tool_use` and one or more `tool_use` content blocks that include:
+19 −19 lines
Recorded here 21 Oct 2026 19:37 UTC
api
agents-and-tools/tool-use/fine-grained-tool-streaming
±changed
+6−6
apichanged
agents-and-tools/tool-use/fine-grained-tool-streaming
1. On `content_block_start` with `type: "tool_use"`, initialize an empty string: `input_json = ""`
2. For each `content_block_delta` with `type: "input_json_delta"`, append: `input_json += event.delta.partial_json`
3. On `content_block_stop`, parse the accumulated string
Guard the parse, as the following SDK examples do. A response can also stop at `max_tokens` midway through a parameter. Check the [stop reason](https://platform.claude.com/docs/en/
+6 −6 lines
Recorded here 21 Oct 2026 19:37 UTC
api
agents-and-tools/tool-use/define-tools
±changed
+21−21
apichanged
agents-and-tools/tool-use/define-tools
| Parameter | Description
| ---------------- | ---------------------------------------------------------------------------------------------------------------------------------------------------------------
| `name` | The name of the tool. Must match the regex `^[a-zA-Z0-9_-]{1,128}$`.
| `description` | A detailed plaintext description of what the tool does, when it should be used, and how it behaves.
| `input_schema` | A [JSON Schema](https://json-schema.org/) object defining the expected parameters for the tool.
+21 −21 lines
Recorded here 21 Oct 2026 19:37 UTC
api
agents-and-tools/tool-use/computer-use-tool
±changed
+20−20
apichanged
agents-and-tools/tool-use/computer-use-tool
The computer use tool is an Anthropic-defined [client toolset](https://platform.claude.com/docs/en/agents-and-tools/tool-use/tool-reference#client-toolsets): one `{"type": "compute
The computer use tool is an Anthropic-defined [client toolset](https://platform.claude.com/docs/en/agents-and-tools/tool-use/tool-reference#client-toolsets): one `{"type": "compute
In some circumstances, Claude will follow commands found in content even when they conflict with your instructions. For example, instructions on webpages or contained in images mig
In some circumstances, Claude follows commands found in content even when they conflict with your instructions. For example, instructions on webpages or contained in images might o
This extra protection won't be ideal for every use case (for example, use cases without a human in the loop), so if you'd like to opt out and turn it off, [contact support](https:/
+20 −20 lines
Recorded here 21 Oct 2026 19:37 UTC
api
agents-and-tools/tool-use/code-execution-tool
±changed
+38−38
apichanged
agents-and-tools/tool-use/code-execution-tool
description: Run Python and bash code in a sandboxed container to analyze data, generate files, and iterate on solutions.
description: Run Python and Bash code in a sandboxed container to analyze data, generate files, and iterate on solutions.
Reach out through the [feedback form](https://forms.gle/LTAU6Xn2puCJMi1n6) to share your feedback on this feature.
Use the [feedback form](https://forms.gle/LTAU6Xn2puCJMi1n6) to share your feedback on this feature.
* `code_execution_20260521` is the same runtime as `code_execution_20260120`. The difference is that the tool description tells Claude about the 90-second wall-clock limit on each
+38 −38 lines
Recorded here 21 Oct 2026 19:37 UTC
api
agents-and-tools/tool-use/build-a-tool-using-agent
±changed
+13−13
apichanged
agents-and-tools/tool-use/build-a-tool-using-agent
model: "claude-opus-5-5",
model: Anthropic::Model::CLAUDE_OPUS_5_5,
model: "claude-opus-5-5",
model: Anthropic::Model::CLAUDE_OPUS_5_5,
The other change is conversation history. Instead of rebuilding the `messages` array from scratch on each request, keep a running list and append to it. Every turn sees the complet
+13 −13 lines
Recorded here 21 Oct 2026 19:37 UTC
api
agents-and-tools/tool-use/browser-use-tool
±changed
+34−34
apichanged
agents-and-tools/tool-use/browser-use-tool
The tool is an Anthropic-defined [client toolset](https://platform.claude.com/docs/en/agents-and-tools/tool-use/tool-reference#client-toolsets): one `browser_toolset_20260801` entr
The tool is an Anthropic-defined [client toolset](https://platform.claude.com/docs/en/agents-and-tools/tool-use/tool-reference#client-toolsets): one `browser_toolset_20260801` entr
Your executor (the part of your application that drives the browser and produces tool results) runs `navigate`, then `read_page`. Your application returns one `tool_result` per blo
Your executor (the part of your application that drives the browser and produces tool results) runs `navigate`, then `read_page`. Your application returns one `tool_result` for eac
* Return one `tool_result` per block in a new `user` message, matched by `tool_use_id`, and echo `"toolset_name": "browser"` on each. Every call must be answered or the next re
+34 −34 lines
Recorded here 21 Oct 2026 19:37 UTC
api
agents-and-tools/tool-use/browser-use-sdk
±changed
+24−24
apichanged
agents-and-tools/tool-use/browser-use-sdk
With the browser and computer toolsets in the Python and TypeScript SDKs, you can build an agent that works in a web browser or on a desktop. It can use applications with no progra
With the browser and computer toolsets in the Python and TypeScript SDKs, you can build an agent that works in a web browser or on a desktop. It can use applications with no progra
Our partners also publish code on GitHub: the [Browser Use quickstart](https://github.com/browser-use/browser-use/tree/main/examples/integrations/toolsets-for-claude), the [Browser
Partner code is also available on GitHub: the [Browser Use quickstart](https://github.com/browser-use/browser-use/tree/main/examples/integrations/toolsets-for-claude), the [Browser
In TypeScript, write members as methods, not arrow-function fields, because the SDK finds them on the prototype. Spell the `type` member `type_` in TypeScript. In Python, it's `typ
+24 −24 lines
Recorded here 21 Oct 2026 19:37 UTC
api
agents-and-tools/tool-use/bash-tool
±changed
+14−14
apichanged
agents-and-tools/tool-use/bash-tool
* **System automation:** Execute scripts, manage files, automate tasks
* **System automation:** Run scripts, manage files, automate tasks
The API is stateless. Nothing about your shell session travels between requests, so your application decides when the session starts, how long it lives, and when to restart it. For
The API is stateless. Nothing about your shell session travels between requests, so your application determines when the session starts, how long it lives, and when to restart it.
The session interleaves stderr with stdout, so error messages land where they happened. The example leaves out what a complete implementation also needs: a timeout that kills t
+14 −14 lines
Recorded here 21 Oct 2026 19:37 UTC
api
agents-and-tools/tool-use/advisor-tool
±changed
+14−14
apichanged
agents-and-tools/tool-use/advisor-tool
The advisor tool lets a faster, lower-cost **executor model** consult a higher-intelligence **advisor model** mid-generation for strategic guidance. The advisor reads the full conv
This pattern fits long-horizon agentic workloads (coding agents, computer use, multistep research pipelines) where most turns are mechanical but having an excellent plan is crucial
The advisor tool lets a faster, lower-cost **executor model** consult a higher-intelligence **advisor model** mid-generation for strategic guidance. The advisor reads the full conv
+14 −14 lines
Recorded here 21 Oct 2026 19:37 UTC
api
agents-and-tools/remote-mcp-servers
±changed
+1−1
apichanged
agents-and-tools/remote-mcp-servers
The remote MCP servers listed below are third-party services designed to work with the Claude API. These servers are not owned, operated, or endorsed by Anthropic. Users should o
The remote MCP servers listed later on this page are third-party services designed to work with the Claude API. These servers are not owned, operated, or endorsed by Anthropic. U
+1 −1 lines
Recorded here 21 Oct 2026 19:37 UTC
api
agents-and-tools/mcp-tunnels/troubleshooting
±changed
+4−4
apichanged
agents-and-tools/mcp-tunnels/troubleshooting
| `curl https://<proxy>:8080` fails with `wrong version number` | Expected; the lis
| `curl https://<proxy>:8080` fails with `wrong version number` | Expected; the lis
OAuth flows fail when your authorization server's source-IP allowlist blocks Anthropic's backend from reaching `/token`, `/register`, and the discovery endpoints. If you'd rather n
OAuth flows fail when your authorization server's source-IP allowlist blocks Anthropic's back end from reaching `/token`, `/register`, and the discovery endpoints. If you'd rather
With this configuration, the user's browser hits `/authorize` on your existing hostname (which your allowlist already permits), while Anthropic's backend reaches `/token`, `/regist
+4 −4 lines
Recorded here 21 Oct 2026 19:37 UTC
api
agents-and-tools/mcp-tunnels/security
±changed
+1−1
apichanged
agents-and-tools/mcp-tunnels/security
* **Require OAuth on every MCP server.** Configure each [upstream MCP server](https://platform.claude.com/docs/en/agents-and-tools/mcp-tunnels/concepts#components) to require OAuth
* **Require OAuth on every MCP server.** Configure each [upstream MCP server](https://platform.claude.com/docs/en/agents-and-tools/mcp-tunnels/concepts#components) to require OAuth
+1 −1 lines
Recorded here 21 Oct 2026 19:37 UTC
api
agents-and-tools/mcp-tunnels/reference
±changed
+3−3
apichanged
agents-and-tools/mcp-tunnels/reference
| `upstream.allowed_ips` | IPv4 CIDR ranges or single addresses the proxy is permitted to connect to. Mutually exclusive with `disable_ip_validation`.
| `upstream.allowed_ips` | IPv4 CIDR ranges or single addresses the proxy is allowed to connect to. Mutually exclusive with `disable_ip_validation`.
* PEM-encoded, single certificate, up to 8 kB.
* PEM-encoded, single certificate, up to 8 kB.
The setup component generates an ECDSA P-256 CA with five-year validity and an RSA 4096-bit server certificate with a wildcard SAN and 90-day validity.
+3 −3 lines
Recorded here 21 Oct 2026 19:37 UTC
api
agents-and-tools/mcp-tunnels/quickstart
±changed
+3−3
apichanged
agents-and-tools/mcp-tunnels/quickstart
This quickstart takes you from zero to Claude calling a private MCP server through a tunnel. It uses Docker Compose with [manual](https://platform.claude.com/docs/en/agents-and-too
This quickstart shows you how to set up a tunnel so that Claude can call a private MCP server through it. It uses Docker Compose with [manual](https://platform.claude.com/docs/en/a
* [OpenSSL](https://openssl-library.org/source/) 1.1.1 or later. Preinstalled on macOS and most Linux distributions; on Windows, install it separately (the `openssl` binary must be
* [OpenSSL](https://openssl-library.org/source/) 1.1.1 or later. Preinstalled on macOS and most Linux distributions. On Windows, install it separately (the `openssl` binary must be
In the Console, go to the [Sessions](https://platform.claude.com/sessions) page (`platform.claude.com/sessions`) and create a session. In the agent picker choose **Create new a
+3 −3 lines
Recorded here 21 Oct 2026 19:37 UTC
api
agents-and-tools/mcp-tunnels/overview
±changed
+9−9
apichanged
agents-and-tools/mcp-tunnels/overview
* **Helm** when deploying to Kubernetes.
* **Docker Compose** for a single host or local testing.
* **Helm** when deploying to Kubernetes
* **Docker Compose** for a single host or local testing
* **Programmatic access** (through Workload Identity Federation) when you have an OIDC identity provider such as a Kubernetes cluster, cloud IAM, or SPIFFE.
+9 −9 lines
Recorded here 21 Oct 2026 19:37 UTC
api
agents-and-tools/mcp-tunnels/deploy-helm
±changed
+8−8
apichanged
agents-and-tools/mcp-tunnels/deploy-helm
The Anthropic Helm chart installs the [tunnel stack](https://platform.claude.com/docs/en/agents-and-tools/mcp-tunnels/concepts#components) as a single Deployment and attaches it to
## Before you begin
The Anthropic Helm chart installs the [tunnel stack](https://platform.claude.com/docs/en/agents-and-tools/mcp-tunnels/concepts#components) as a single Deployment and attaches it to
+8 −8 lines
Recorded here 21 Oct 2026 19:37 UTC
api
agents-and-tools/mcp-tunnels/deploy-compose
±changed
+4−4
apichanged
agents-and-tools/mcp-tunnels/deploy-compose
## Before you begin
## Before you start
* **A tunnel.** With programmatic access, the [setup component](https://platform.claude.com/docs/en/agents-and-tools/mcp-tunnels/concepts#components) creates one for you when you d
* **A tunnel.** With programmatic access, the [setup component](https://platform.claude.com/docs/en/agents-and-tools/mcp-tunnels/concepts#components) creates one for you when you d
The proxy listens on `:8080` over plain WebSocket; the [inner TLS](https://platform.claude.com/docs/en/agents-and-tools/mcp-tunnels/concepts#components) handshake happens *
+4 −4 lines
Recorded here 21 Oct 2026 19:37 UTC
api
agents-and-tools/mcp-tunnels/console
±changed
+11−11
apichanged
agents-and-tools/mcp-tunnels/console
title: Manage tunnels in the Console
title: Manage tunnels in the Claude Console
Click **New tunnel** and enter a name in the **Create tunnel** dialog. The name is required and identifies the tunnel in the list, on the detail page, and in the agent MCP serv
Click **New tunnel** and enter a name in the **Create tunnel** dialog box. The name is required and identifies the tunnel in the list, on the detail page, and in the agent MCP
2. **A federation rule with the `workspace:manage_tunnels` scope.** Turning on the toggle reveals a **Federation rule** picker. Choose an existing rule with that scope, or clic
+11 −11 lines
Recorded here 21 Oct 2026 19:37 UTC
api
agents-and-tools/mcp-tunnels/concepts
±changed
+13−13
apichanged
agents-and-tools/mcp-tunnels/concepts
This page defines the terms used throughout the [MCP tunnels](https://platform.claude.com/docs/en/agents-and-tools/mcp-tunnels/overview) documentation. Several components appear un
This page defines the terms used throughout the [MCP tunnels](https://platform.claude.com/docs/en/agents-and-tools/mcp-tunnels/overview) documentation. Several components appear un
| Term | Definition
| ----------------------- | --------------------------------------------------------------------------------------------------------------------------------------------------------
| **Tunnel stack** | The two containers you run inside your network to attach to a tunnel: the proxy and cloudflared. One stack serves one tunnel and can be replicated acros
+13 −13 lines
Recorded here 21 Oct 2026 19:37 UTC
api
agents-and-tools/mcp-connector
±changed
+18−18
apichanged
agents-and-tools/mcp-connector
Claude's Model Context Protocol (MCP) connector feature enables you to connect to remote MCP servers directly from the Messages API without a separate MCP client.
Claude's Model Context Protocol (MCP) connector feature allows you to connect to remote MCP servers directly from the Messages API without a separate MCP client.
| `configs` | object | No | Per-tool configuration overrides. Keys are tool names, values are configuration objects. |
| `configs` | object | No | Per-tool configuration overrides. Keys are tool names; values are configuration objects. |
* `search_events`: `enabled: false` (from configs), `defer_loading: true` (from default\_config)
+18 −18 lines
Recorded here 21 Oct 2026 19:37 UTC
api
agents-and-tools/agent-skills/quickstart
±changed
+1−1
apichanged
agents-and-tools/agent-skills/quickstart
First, check what Skills are available. Use the Skills API to list all Anthropic-managed Skills. Each language tab is an excerpt from one continuous script, with any imports and cl
First, check what Skills are available. Use the Skills API to list all Anthropic-managed Skills. The example is an excerpt from one continuous script, with any imports and client s
+1 −1 lines
Recorded here 21 Oct 2026 19:37 UTC
api
agents-and-tools/agent-skills/overview
±changed
+33−33
apichanged
agents-and-tools/agent-skills/overview
This filesystem-based architecture enables **progressive disclosure:** Claude loads information in stages as needed, rather than consuming context upfront.
This filesystem-based architecture allows **progressive disclosure:** Claude loads information in stages as needed, rather than consuming context upfront.
The main body of SKILL.md contains procedural knowledge: workflows, best practices, and guidance:
The main body of `SKILL.md` contains procedural knowledge: workflows, best practices, and guidance:
When you request something that matches a Skill's description, Claude reads SKILL.md from the filesystem using bash. Only then does this content enter the context window.
+33 −33 lines
Recorded here 21 Oct 2026 19:37 UTC
api
agents-and-tools/agent-skills/enterprise
±changed
+17−17
apichanged
agents-and-tools/agent-skills/enterprise
This guide is for enterprise admins and architects who need to govern Agent Skills across an organization. It covers how to vet, evaluate, deploy, and manage Skills at scale. For a
This guide is for enterprise admins and architects who need to govern Agent Skills across an organization. It covers how to vet, evaluate, deploy, and manage Skills at scale. For a
| Filesystem access scope | Paths outside the Skill directory, broad glob patterns, path traversal (`../`) | Medium: may access unintended data
| Filesystem access scope | Paths outside the Skill directory, broad glob patterns, path traversal (`../`) | Medium: might access unintended data
1. **Read all Skill directory content.** Review SKILL.md, all referenced markdown files, and any bundled scripts or resources.
+17 −17 lines
Recorded here 21 Oct 2026 19:37 UTC
api
agents-and-tools/agent-skills/claude-api-skill
±changed
+10−10
apichanged
agents-and-tools/agent-skills/claude-api-skill
* **Messages API:** The primary surface for single requests, streaming chat, tool use, batch processing, prompt caching, structured outputs, and custom agent loops.
* **Claude Managed Agents (beta):** An Anthropic-hosted surface for server-managed stateful agents with Anthropic-hosted tool execution, persistent agent configs, and per-session s
* **Messages API:** The primary surface for single requests, streaming chat, tool use, batch processing, prompt caching, structured outputs, and custom agent loops
* **Claude Managed Agents (beta):** An Anthropic-hosted surface for server-managed stateful agents with Anthropic-hosted tool execution, persistent agent configs, and per-session s
* **Batch processing:** Offline batch processing at 50% cost
+10 −10 lines
Recorded here 21 Oct 2026 19:37 UTC
api
agents-and-tools/agent-skills/best-practices
±changed
+83−83
apichanged
agents-and-tools/agent-skills/best-practices
For conceptual background on how Skills work, see the [Skills overview](https://platform.claude.com/docs/en/agents-and-tools/agent-skills/overview).
For conceptual background on how Skills work, see [Agent Skills](https://platform.claude.com/docs/en/agents-and-tools/agent-skills/overview).
The [context window](https://platform.claude.com/docs/en/build-with-claude/context-windows) is a public good. Your Skill shares the context window with everything else Claude needs
The [context window](https://platform.claude.com/docs/en/build-with-claude/context-windows) is a public good. Your Skill shares the context window with everything else Claude needs
Not every token in your Skill has an immediate cost. At startup, only the metadata (name and description) from all Skills is pre-loaded. Claude reads SKILL.md only when the Skill b
+83 −83 lines
Recorded here 21 Oct 2026 19:37 UTC
api
about-claude/use-case-guides/ticket-routing
±changed
+43−43
apichanged
about-claude/use-case-guides/ticket-routing
description: This guide walks through how to harness Claude's advanced natural language understanding capabilities to classify customer support tickets at scale based on customer i
description: This guide walks through how to use Claude's advanced natural language understanding capabilities to classify customer support tickets at scale based on customer inten
Here are some key indicators that you should use an LLM like Claude instead of traditional ML approaches for your classification task:
Here are some key indicators that you should use an LLM such as Claude instead of traditional ML approaches for your classification task:
<Accordion title="You require interpretable reasoning for classification decisions">
+43 −43 lines
Recorded here 21 Oct 2026 19:37 UTC
api
about-claude/use-case-guides/legal-summarization
±changed
+24−24
apichanged
about-claude/use-case-guides/legal-summarization
description: This guide walks through how to leverage Claude's advanced natural language processing capabilities to efficiently summarize legal documents, extracting key informatio
description: This guide walks through how to use Claude's advanced natural language processing capabilities to efficiently summarize legal documents, extracting key information and
> Visit the [summarization cookbook](https://platform.claude.com/cookbook/capabilities-summarization-guide) to see an example legal summarization implementation using Claude.
> Visit the [summarization recipe](https://platform.claude.com/cookbook/capabilities-summarization-guide) in the Claude Cookbook to see an example legal summarization implementatio
Large-scale document review can be time-consuming and expensive when done manually. Claude can process and summarize vast amounts of legal documents rapidly, significantly redu
+24 −24 lines
Recorded here 21 Oct 2026 19:37 UTC
api
about-claude/use-case-guides/customer-support-chat
±changed
+42−42
apichanged
about-claude/use-case-guides/customer-support-chat
Here are some key indicators that you should employ an LLM like Claude to automate portions of your customer support process:
Here are some key indicators that you should employ an LLM such as Claude to automate portions of your customer support process:
Claude can quickly retrieve, process, and combine information from vast knowledge bases, while human agents may need time to research or consult multiple sources.
Claude can quickly retrieve, process, and combine information from vast knowledge bases, whereas human agents might need time to research or consult multiple sources.
Claude can provide round-the-clock support without fatigue, whereas staffing human agents for continuous coverage can be costly and challenging.
+42 −42 lines
Recorded here 21 Oct 2026 19:37 UTC
api
about-claude/use-case-guides/content-moderation
±changed
+21−21
apichanged
about-claude/use-case-guides/content-moderation
> Visit the [content moderation cookbook](https://platform.claude.com/cookbook/misc-building-moderation-filter) to see an example content moderation implementation using Claude.
> Visit the [content moderation recipe](https://platform.claude.com/cookbook/misc-building-moderation-filter) in the Claude Cookbook to see an example content moderation implementa
Here are some key indicators that you should use an LLM like Claude instead of a traditional ML or rules-based approach for content moderation:
Here are some key indicators that you should use an LLM such as Claude instead of a traditional ML or rules-based approach for content moderation:
Traditional ML approaches, such as bag-of-words models or simple pattern matching, often struggle to understand the tone, intent, and context of the content. While human modera
+21 −21 lines
Recorded here 21 Oct 2026 19:37 UTC
api
about-claude/pricing
±changed
+22−22
apichanged
about-claude/pricing
Regional and multi-region endpoints include a 10% premium over global endpoints. The Claude API (first-party) is global by default; for first-party data residency options and pri
Regional and multi-region endpoints include a 10 percent premium over global endpoints. The Claude API (first-party) is global by default; for first-party data residency options
[Claude Platform on AWS](https://platform.claude.com/docs/en/build-with-claude/claude-platform-on-aws) bills through AWS Marketplace using Claude Consumption Units (CCUs). Anthropi
[Claude Platform on AWS](https://platform.claude.com/docs/en/build-with-claude/claude-platform-on-aws) bills through AWS Marketplace using Claude Consumption Units (CCUs). Anthropi
[Claude in Microsoft Foundry](https://platform.claude.com/docs/en/build-with-claude/claude-in-microsoft-foundry) bills through the Azure Marketplace using Claude Consumption Units
+22 −22 lines
Recorded here 21 Oct 2026 19:37 UTC
api
about-claude/models/optimizing-for-cost-and-intelligence
±changed
+15−15
apichanged
about-claude/models/optimizing-for-cost-and-intelligence
Prompt caching, token hygiene, batch processing, and a prompt audit against your current model all lower what you pay without lowering output quality. Two caveats apply: batch proc
Prompt caching, token hygiene, batch processing, and a prompt audit against your current model all lower what you pay without lowering output quality. Two caveats apply: batch proc
3. Send it within 4 minutes of the previous request's start, and every 4 minutes after that. Count from the request's start, because the [cache's 5-minute lifetime](https://platfor
3. Send it within 4 minutes of the previous request's start, and every 4 minutes after that, until the next turn arrives or the session ends. Count from the request's start, becaus
Pasted into the prompt, the table is about 91,000 input tokens on every request, and Claude Sonnet 5 answered 7 of 25 questions correctly. Uploaded, with code execution, it answere
+15 −15 lines
Recorded here 21 Oct 2026 19:37 UTC
api
about-claude/models/choosing-a-model
±changed
+6−6
apichanged
about-claude/models/choosing-a-model
For many applications, starting with a faster, more cost-effective model like Claude Haiku 5.5 can be the optimal approach:
For many applications, starting with a faster, more cost-effective model such as Claude Haiku 5.5 can be the optimal approach:
1. Begin implementation with Claude Haiku 5.5.
1. Start implementation with Claude Haiku 5.5.
This approach allows for quick iteration, lower development costs, and is often sufficient for many common applications. This approach is best for:
+6 −6 lines
Recorded here 21 Oct 2026 19:37 UTC
api
about-claude/model-deprecations
±changed
+14−14
apichanged
about-claude/model-deprecations
* **Legacy:** The model will no longer receive updates and may be deprecated in the future.
* **Legacy:** The model will no longer receive updates and might be deprecated in the future.
* **Retired:** The model is no longer available for use. Requests to retired models will fail.
* **Retired:** The model is no longer available for use. Requests to retired models fail.
For specific instructions on migrating to the latest Claude models, see the [Migration guide](https://platform.claude.com/docs/en/about-claude/models/migration-guide).
+14 −14 lines
Recorded here 21 Oct 2026 19:37 UTC
api
about-claude/glossary
±changed
+11−11
apichanged
about-claude/glossary
The "context window" refers to the amount of text a language model can look back on and reference when generating new text. This is different from the large corpus of data the lang
The "context window" refers to the amount of text a language model can look back on and reference when generating new text. This is different from the large corpus of data the lang
Model Context Protocol (MCP) is an open protocol that standardizes how applications provide context to LLMs. Like a USB-C port for AI applications, MCP provides a unified way to co
Model Context Protocol (MCP) is an open protocol that standardizes how applications provide context to LLMs. Like a USB-C port for AI applications, MCP provides a unified way to co
The MCP connector is a feature that allows API users to connect to MCP servers directly from the Messages API without building an MCP client. This enables seamless integration with
+11 −11 lines
Recorded here 21 Oct 2026 19:37 UTC
api
about-claude/api-credits-for-subscribers
±changed
+8−8
apichanged
about-claude/api-credits-for-subscribers
## At a glance
## Summary
## API credits amounts
## API credit amounts
### Before you begin
+8 −8 lines
Recorded here 21 Oct 2026 19:37 UTC
api
about-claude/additional-resources
±changed
+1−1
apichanged
about-claude/additional-resources
<Card title="Cookbook" icon="fork-knife" href="https://platform.claude.com/cookbook">
<Card title="Claude Cookbook" icon="fork-knife" href="https://platform.claude.com/cookbook">
+1 −1 lines
Recorded here 21 Oct 2026 19:37 UTC
api
test-and-evaluate/strengthen-guardrails/reduce-latency
±changed
+0−1
apichanged
test-and-evaluate/strengthen-guardrails/reduce-latency
* **Experiment with temperature:** The `temperature` [parameter](https://platform.claude.com/docs/en/api/messages/create) controls the randomness of the output. Lower values (for e
+0 −1 lines
Recorded here 21 Oct 2026 18:37 UTC
api
release-notes/overview
±changed
+6−0
apichanged
release-notes/overview
### October 21, 2026
* We've released API version `2026-10-21`. [Migrating to API version 2026-10-21](https://platform.claude.com/docs/en/api/migration-2026-10-21) shows how to upgrade.
* We've released new versions of the [client SDKs](https://platform.claude.com/docs/en/cli-sdks-libraries/overview) and the [`ant` CLI](https://platform.claude.com/docs/en/cli-sdks
+6 −0 lines
Recorded here 21 Oct 2026 18:37 UTC
api
models/sonnet-5/overview
±changed
+1−1
apichanged
models/sonnet-5/overview
* Setting `temperature`, `top_p`, or `top_k` to non-default values returns a 400 error.
* Setting `temperature`, `top_p`, or `top_k` returns a 400 error.
+1 −1 lines
Recorded here 21 Oct 2026 18:37 UTC
api
models/sonnet-5-5/overview
±changed
+1−1
apichanged
models/sonnet-5-5/overview
* Setting `temperature`, `top_p`, or `top_k` to a non-default value returns a 400 error.
* Setting `temperature`, `top_p`, or `top_k` returns a 400 error.
+1 −1 lines
Recorded here 21 Oct 2026 18:37 UTC
Nothing on this page matches that. Press Enter to search every page of the ledger.