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
251-275 of 306, page 11 of 13This capture is too large to show at once. Changes 251-275 of 306 are below, significant first; the rest are on the following screens.
docs/2025-11-25/develop/clients/client-best-practices Changed · +2 / -2 lines
from line 10
1010
1111Naive MCP host implementations pass the tool definitions of every connected server directly to the model at the start of each conversation. For a handful of tools, this is perfectly reasonable. But when a host has access to dozens of servers exposing hundreds of tools, those definitions alone can consume the majority of the context window before the model has even read the user's message.
1212
13<img src="https://mintcdn.com/mcp/JXfd5cBmEUh_qPUI/images/progressive-discovery.svg?fit=max&auto=format&n=JXfd5cBmEUh_qPUI&q=85&s=db39f47006107f04af43b5eeae2d6022" alt="Comparison of loading all tools upfront versus discovering tools on demand. The upfront approach consumes ~150,000 tokens on definitions alone, while progressive discovery uses ~2,000 tokens by loading only what the task requires." width="760" height="440" data-path="images/progressive-discovery.svg" />
13
1414
1515Progressive discovery avoids this:
1616
from line 156
156156Programmatic tool calling is powerful and allows for more efficient use of MCP tools and resources, but requires
157157clients to implement a sandbox environment.
158158
159<img src="https://mintcdn.com/mcp/JXfd5cBmEUh_qPUI/images/programmatic-tool-calling.svg?fit=max&auto=format&n=JXfd5cBmEUh_qPUI&q=85&s=a2be82d097bb7cd7c7fd415918b1571d" alt="Comparison of direct tool calling versus programmatic tool calling. Direct calling passes every intermediate result through the model (~100K+ tokens). Programmatic calling sends a ~200-token script to a sandbox, which executes the tool calls and returns a ~15-token summary." width="900" height="900" data-path="images/programmatic-tool-calling.svg" />
159
160160
161161### How It Works
162162
docs/2026-07-28/develop/clients/client-best-practices Changed · +2 / -2 lines
from line 10
1010
1111Naive MCP host implementations pass the tool definitions of every connected server directly to the model at the start of each conversation. For a handful of tools, this is perfectly reasonable. But when a host has access to dozens of servers exposing hundreds of tools, those definitions alone can consume the majority of the context window before the model has even read the user's message.
1212
13<img src="https://mintcdn.com/mcp/JXfd5cBmEUh_qPUI/images/progressive-discovery.svg?fit=max&auto=format&n=JXfd5cBmEUh_qPUI&q=85&s=db39f47006107f04af43b5eeae2d6022" alt="Comparison of loading all tools upfront versus discovering tools on demand. The upfront approach consumes ~150,000 tokens on definitions alone, while progressive discovery uses ~2,000 tokens by loading only what the task requires." width="760" height="440" data-path="images/progressive-discovery.svg" />
13
1414
1515Progressive discovery avoids this:
1616
from line 165
165165Programmatic tool calling is powerful and allows for more efficient use of MCP tools and resources, but requires
166166clients to implement a sandbox environment.
167167
168<img src="https://mintcdn.com/mcp/JXfd5cBmEUh_qPUI/images/programmatic-tool-calling.svg?fit=max&auto=format&n=JXfd5cBmEUh_qPUI&q=85&s=a2be82d097bb7cd7c7fd415918b1571d" alt="Comparison of direct tool calling versus programmatic tool calling. Direct calling passes every intermediate result through the model (~100K+ tokens). Programmatic calling sends a ~200-token script to a sandbox, which executes the tool calls and returns a ~15-token summary." width="900" height="900" data-path="images/programmatic-tool-calling.svg" />
168
169169
170170### How It Works
171171
docs/draft/develop/clients/client-best-practices Changed · +2 / -2 lines
from line 10
1010
1111Naive MCP host implementations pass the tool definitions of every connected server directly to the model at the start of each conversation. For a handful of tools, this is perfectly reasonable. But when a host has access to dozens of servers exposing hundreds of tools, those definitions alone can consume the majority of the context window before the model has even read the user's message.
1212
13<img src="https://mintcdn.com/mcp/JXfd5cBmEUh_qPUI/images/progressive-discovery.svg?fit=max&auto=format&n=JXfd5cBmEUh_qPUI&q=85&s=db39f47006107f04af43b5eeae2d6022" alt="Comparison of loading all tools upfront versus discovering tools on demand. The upfront approach consumes ~150,000 tokens on definitions alone, while progressive discovery uses ~2,000 tokens by loading only what the task requires." width="760" height="440" data-path="images/progressive-discovery.svg" />
13
1414
1515Progressive discovery avoids this:
1616
from line 165
165165Programmatic tool calling is powerful and allows for more efficient use of MCP tools and resources, but requires
166166clients to implement a sandbox environment.
167167
168<img src="https://mintcdn.com/mcp/JXfd5cBmEUh_qPUI/images/programmatic-tool-calling.svg?fit=max&auto=format&n=JXfd5cBmEUh_qPUI&q=85&s=a2be82d097bb7cd7c7fd415918b1571d" alt="Comparison of direct tool calling versus programmatic tool calling. Direct calling passes every intermediate result through the model (~100K+ tokens). Programmatic calling sends a ~200-token script to a sandbox, which executes the tool calls and returns a ~15-token summary." width="900" height="900" data-path="images/programmatic-tool-calling.svg" />
168
169169
170170### How It Works
171171
specification/2024-11-05/server/prompts Changed · +1 / -1 lines
from line 15
1515
1616For example, as slash commands:
1717
18<img src="https://mintcdn.com/mcp/4ZXF1PrDkEaJvXpn/specification/2024-11-05/server/slash-command.png?fit=max&auto=format&n=4ZXF1PrDkEaJvXpn&q=85&s=ae78d658be66d75446ffd37a36e944c9" alt="Example of prompt exposed as slash command" width="293" height="106" data-path="specification/2024-11-05/server/slash-command.png" />
18
1919
2020However, implementors are free to expose prompts through any interface pattern that suits
2121their needs—the protocol itself does not mandate any specific user interaction
specification/2024-11-05/server/resources Changed · +1 / -1 lines
from line 17
1717* Allow the user to search through and filter available resources
1818* Implement automatic context inclusion, based on heuristics or the AI model's selection
1919
20<img src="https://mintcdn.com/mcp/4ZXF1PrDkEaJvXpn/specification/2024-11-05/server/resource-picker.png?fit=max&auto=format&n=4ZXF1PrDkEaJvXpn&q=85&s=c393fd9aa41eca521cb2f830d839612f" alt="Example of resource context picker" width="174" height="181" data-path="specification/2024-11-05/server/resource-picker.png" />
20
2121
2222However, implementations are free to expose resources through any interface pattern that
2323suits their needs—the protocol itself does not mandate any specific user
specification/2025-03-26/server/prompts Changed · +1 / -1 lines
from line 15
1515
1616For example, as slash commands:
1717
18<img src="https://mintcdn.com/mcp/4ZXF1PrDkEaJvXpn/specification/2025-03-26/server/slash-command.png?fit=max&auto=format&n=4ZXF1PrDkEaJvXpn&q=85&s=8390583e2400c914dd404e37e014613e" alt="Example of prompt exposed as slash command" width="293" height="106" data-path="specification/2025-03-26/server/slash-command.png" />
18
1919
2020However, implementors are free to expose prompts through any interface pattern that suits
2121their needs—the protocol itself does not mandate any specific user interaction
specification/2025-03-26/server/resources Changed · +1 / -1 lines
from line 17
1717* Allow the user to search through and filter available resources
1818* Implement automatic context inclusion, based on heuristics or the AI model's selection
1919
20<img src="https://mintcdn.com/mcp/4ZXF1PrDkEaJvXpn/specification/2025-03-26/server/resource-picker.png?fit=max&auto=format&n=4ZXF1PrDkEaJvXpn&q=85&s=7f6d9a491a97e714b5d5fc74cb0c7132" alt="Example of resource context picker" width="174" height="181" data-path="specification/2025-03-26/server/resource-picker.png" />
20
2121
2222However, implementations are free to expose resources through any interface pattern that
2323suits their needs—the protocol itself does not mandate any specific user
specification/2025-06-18/architecture/index Changed · +0 / -2 lines
from line 1
11# Architecture
22
3<div id="enable-section-numbers" />
4
53The Model Context Protocol (MCP) follows a client-host-server architecture where each
64host can run multiple client instances. This architecture enables users to integrate AI
75capabilities across applications while maintaining clear security boundaries and
specification/2025-06-18/basic/authorization Changed · +0 / -2 lines
from line 1
11# Authorization
22
3<div id="enable-section-numbers" />
4
53## Introduction
64
75### Purpose and Scope
specification/2025-06-18/basic/index Changed · +0 / -2 lines
from line 1
11# Overview
22
3<div id="enable-section-numbers" />
4
53The Model Context Protocol consists of several key components that work together:
64
75* **Base Protocol**: Core JSON-RPC message types
specification/2025-06-18/basic/utilities/cancellation Changed · +0 / -2 lines
from line 1
11# Cancellation
22
3<div id="enable-section-numbers" />
4
53The Model Context Protocol (MCP) supports optional cancellation of in-progress requests
64through notification messages. Either side can send a cancellation notification to
75indicate that a previously-issued request should be terminated.
specification/2025-06-18/basic/utilities/ping Changed · +0 / -2 lines
from line 1
11# Ping
22
3<div id="enable-section-numbers" />
4
53The Model Context Protocol includes an optional ping mechanism that allows either party
64to verify that their counterpart is still responsive and the connection is alive.
75
specification/2025-06-18/basic/utilities/progress Changed · +0 / -2 lines
from line 1
11# Progress
22
3<div id="enable-section-numbers" />
4
53The Model Context Protocol (MCP) supports optional progress tracking for long-running
64operations through notification messages. Either side can send progress notifications to
75provide updates about operation status.
specification/2025-06-18/changelog Changed · +0 / -2 lines
from line 1
11# Key Changes
22
3<div id="enable-section-numbers" />
4
53This document lists changes made to the Model Context Protocol (MCP) specification since
64the previous revision, [2025-03-26](/specification/2025-03-26).
75
specification/2025-06-18/client/roots Changed · +0 / -2 lines
from line 1
11# Roots
22
3<div id="enable-section-numbers" />
4
53The Model Context Protocol (MCP) provides a standardized way for clients to expose
64filesystem "roots" to servers. Roots define the boundaries of where servers can operate
75within the filesystem, allowing them to understand which directories and files they have
specification/2025-06-18/server/resources Changed · +1 / -3 lines
from line 1
11# Resources
22
3<div id="enable-section-numbers" />
4
53The Model Context Protocol (MCP) provides a standardized way for servers to expose
64resources to clients. Resources allow servers to share data that provides context to
75language models, such as files, database schemas, or application-specific information.
from line 17
1917* Allow the user to search through and filter available resources
2018* Implement automatic context inclusion, based on heuristics or the AI model's selection
2119
22<img src="https://mintcdn.com/mcp/4ZXF1PrDkEaJvXpn/specification/2025-06-18/server/resource-picker.png?fit=max&auto=format&n=4ZXF1PrDkEaJvXpn&q=85&s=133fa885ef6e9c2e20049da5c33f4386" alt="Example of resource context picker" width="174" height="181" data-path="specification/2025-06-18/server/resource-picker.png" />
20
2321
2422However, implementations are free to expose resources through any interface pattern that
2523suits their needs—the protocol itself does not mandate any specific user
specification/2025-06-18/server/utilities/completion Changed · +0 / -2 lines
from line 1
11# Completion
22
3<div id="enable-section-numbers" />
4
53The Model Context Protocol (MCP) provides a standardized way for servers to offer
64argument autocompletion suggestions for prompts and resource URIs. This enables rich,
75IDE-like experiences where users receive contextual suggestions while entering argument
specification/2025-06-18/server/utilities/logging Changed · +0 / -2 lines
from line 1
11# Logging
22
3<div id="enable-section-numbers" />
4
53The Model Context Protocol (MCP) provides a standardized way for servers to send
64structured log messages to clients. Clients can control logging verbosity by setting
75minimum log levels, with servers sending notifications containing severity levels,
specification/2025-06-18/server/utilities/pagination Changed · +0 / -2 lines
from line 1
11# Pagination
22
3<div id="enable-section-numbers" />
4
53The Model Context Protocol (MCP) supports paginating list operations that may return
64large result sets. Pagination allows servers to yield results in smaller chunks rather
75than all at once.
specification/2025-11-25/architecture/index Changed · +0 / -2 lines
from line 1
11# Architecture
22
3<div id="enable-section-numbers" />
4
53The Model Context Protocol (MCP) follows a client-host-server architecture where each
64host can run multiple client instances. This architecture enables users to integrate AI
75capabilities across applications while maintaining clear security boundaries and
specification/2025-11-25/basic/authorization Changed · +0 / -2 lines
from line 1
11# Authorization
22
3<div id="enable-section-numbers" />
4
53## Introduction
64
75### Purpose and Scope
specification/2025-11-25/basic/index Changed · +0 / -2 lines
from line 1
11# Overview
22
3<div id="enable-section-numbers" />
4
53The Model Context Protocol consists of several key components that work together:
64
75* **Base Protocol**: Core JSON-RPC message types
specification/2025-11-25/basic/utilities/cancellation Changed · +0 / -2 lines
from line 1
11# Cancellation
22
3<div id="enable-section-numbers" />
4
53The Model Context Protocol (MCP) supports optional cancellation of in-progress requests
64through notification messages. Either side can send a cancellation notification to
75indicate that a previously-issued request should be terminated.
specification/2025-11-25/basic/utilities/ping Changed · +0 / -2 lines
from line 1
11# Ping
22
3<div id="enable-section-numbers" />
4
53The Model Context Protocol includes an optional ping mechanism that allows either party
64to verify that their counterpart is still responsive and the connection is alive.
75
specification/2025-11-25/basic/utilities/progress Changed · +0 / -2 lines
from line 1
11# Progress
22
3<div id="enable-section-numbers" />
4
53The Model Context Protocol (MCP) supports optional progress tracking for long-running
64operations through notification messages. Either side can send progress notifications to
75provide updates about operation status.