Model Context Protocol (MCP)
An open standard from Anthropic that connects AI agents to tools and data through a single JSON-RPC interface — the de facto agent-to-tool standard.
- term
- Model Context Protocol (MCP)
- category
- protocols
- short_def
- An open standard from Anthropic that connects AI agents to tools and data through a single JSON-RPC interface — the de facto agent-to-tool standard.
- long_def
- MCP standardizes how an agent discovers and calls tools, reads resources and uses prompts, so any compatible agent can talk to any compatible server without bespoke integration. Introduced by Anthropic in November 2024 and now governed under the Linux Foundation via the AAIF (formed December 2025).
- see_also
webmcptool-useai-agent- etymology_origin
- Introduced and open-sourced by Anthropic on 25 November 2024; governance moved to the Linux Foundation's Agentic AI Infrastructure Foundation (AAIF), formed December 2025.
- related_to
webmcptool-useai-agenta2anlweb- contrast_with
- Unlike A2A, which connects agents to other agents, MCP connects a single agent to tools and data — they complement rather than compete, the most common confusion in the vocabulary.
- example
- Within a year of its November 2024 launch, MCP was adopted across the ecosystem and is housed alongside A2A under the AAIF (Linux Foundation, formed December 2025).
- source
- https://en.wikipedia.org/wiki/Model_Context_Protocol
- status
- active
- why_it_matters
- MCP is the dominant way a website exposes callable tools to agents; implementing or speaking MCP is a core agent-readiness capability.
- sameAs
https://en.wikipedia.org/wiki/Model_Context_Protocolhttps://modelcontextprotocol.io/- bridge_entity
- protocols/capability/mcp
- last_verified
- 2026-06-15
- md_twin
- /glossary/mcp.md