Model Context Protocol (MCP)
An open protocol, built on JSON-RPC, for connecting LLM applications to external data sources and tools through hosts, clients and servers.
Canonical text: https://modelcontextprotocol.io/specification/2026-07-28, read against version 2026-07-28.
Key principles
- Users must explicitly consent to and understand all data access and operations.
- Users must retain control over what data is shared and what actions are taken.
- Implementors should provide clear UIs for reviewing and authorizing activities.
- Hosts must obtain explicit user consent before exposing user data to servers.
- Hosts must not transmit resource data elsewhere without user consent.
- User data should be protected with appropriate access controls.
- Tools represent arbitrary code execution and must be treated with appropriate caution.
- Hosts must obtain explicit user consent before invoking any tool.
- Users should understand what each tool does before authorizing its use.
- Base protocol: JSON-RPC message format, stateless self-contained requests, per-request capability negotiation.
- Servers offer Resources, Prompts and Tools; clients may offer Elicitation.
- Extensions are always opt-in and require explicit support from both client and server.
Facts
- Maintainer
- MCP Steering Group, Model Context Protocol a Series of LF Projects, LLC
- Repository
- https://github.com/modelcontextprotocol/modelcontextprotocol
- Licence of the text
- Apache 2.0
- First published
- 2024-11-05
- Status of the text
- stable
- Listing
- listed, not yet claimed
Read with
- claimed
12-Factor AI-enabled CLI
Twelve testable factors for a command-line tool whose operator may be a human or an AI agent, from machine-readable capabilities to meaningful exit codes, with a conformance gate.
v0.1 Agentic Stack - claimed
Purpose-bound Access Control and Trust
How an AI agent proves who it acts for, with what declared purpose and evidence, to a gateway that already authenticates and authorizes; a PACT pass never grants access and a native denial always wins.
v1.0.0-alpha.0 Agentic Stack -
Agent2Agent Protocol
An open protocol for communication and interoperability between opaque agentic applications, with JSON-RPC, gRPC and HTTP+JSON bindings and an Agent Card for discovery.
v1.0.0 Agentic Stack -
Agentic Permissions Policy Algebra
A declarative policy format and deterministic engine that labels data flowing between an LLM agent and its tools, so restricted data cannot reach audiences it was never allowed to reach.
vv0.37.0 Agentic Stack
Badge
[](https://manifesto.md/mcp/) Maintainers claim a listing with a pull request setting listing: claimed, or a yes on the listing issue.