# Model Context Protocol vs Agent Client Protocol

The map lists them as complementary: sites and agents commonly implement both.

[5 Act](/category/act/) [Model Context Protocol](/protocols/mcp/) · [5 Act](/category/act/) [Agent Client Protocol](/protocols/agent-client-protocol/)

|  | [Model Context Protocol](/protocols/mcp/) MCP | [Agent Client Protocol](/protocols/agent-client-protocol/) ACP |
| --- | --- | --- |
| Lifecycle step | 5 Act: Do something for me | 5 Act: Do something for me |
| Answers | How does an AI application call tools and pull structured context, prompts, and data from an external server? | How does a code editor invoke and converse with an interchangeable coding agent it did not build? |
| Governed by | Agentic AI Foundation (Linux Foundation umbrella); day-to-day spec work in the modelcontextprotocol GitHub org via public SEPs (Specification Enhancement Proposals) and working groups | Zed Industries acts as BDFL/design-team lead over a public Request for Dialog (RFD) process; Sergey Ignatov (JetBrains) joined as a second Lead Maintainer on 2026-02-18, reflecting growing Zed/JetBrains collaboration |
| Status (issuer's words) | Active open specification, current dated release 2026-07-28 | v1 stable and shipping; v2 published in Draft form as of 2026-07-20 |
| Phase | living | stable |
| Version | 2026-07-28 (2026-07-28) | v1.11.0 (schema v1.25.0); v2 unstable (2026-10-09) |
| Form / transport | endpoint | endpoint |
| Found at | GET {endpoint}/server-card (SEP-2127 Server Card, Final 2026-10-06; found via an AI Catalog entry) | none fixed; local agents are launched as sub-processes, remote agents connect over HTTP or WebSocket (remote support still a work in progress) |
| Adoption | claimed — Wikipedia (OpenAI, Google DeepMind adopted MCP by March 2025) and WorkOS ('adopted by OpenAI, Google DeepMind, Microsoft, and thousands of development teams') | unmeasured |
| Evidence | 7 verified, 2 reported | 4 verified, 1 reported |
| Primary use case | Expose tools to agents | Describe an agent's capabilities |
| Last verified | 2026-10-19 (current) | 2026-10-19 (current) |
| Specification | [https://modelcontextprotocol.io/specification/2026-07-28/index](https://modelcontextprotocol.io/specification/2026-07-28/index) | [https://agentclientprotocol.com/get-started/introduction](https://agentclientprotocol.com/get-started/introduction) |

## When to use which

- Choose Model Context Protocol when your question is “How does an AI application call tools and pull structured context, prompts, and data from an external server?” and you accept its maturity: phase living, adoption claimed (Wikipedia (OpenAI, Google DeepMind adopted MCP by March 2025) and WorkOS ('adopted by OpenAI, Google DeepMind, Microsoft, and thousands of development teams')).
- Choose Agent Client Protocol when your question is “How does a code editor invoke and converse with an interchangeable coding agent it did not build?” and you accept its maturity: phase stable, adoption unmeasured.

Guidance is derived from the two records (question, phase, adoption, prerequisites, pairings), not from opinion. Each record's page carries its numbered references.

## Read the full articles

- [Model Context Protocol](/protocols/mcp/): How does an AI application call tools and pull structured context, prompts, and data from an external server?
- [Agent Client Protocol](/protocols/agent-client-protocol/): How does a code editor invoke and converse with an interchangeable coding agent it did not build?

Ask the MCP server to compare any two: the compare-protocols prompt, or get on both ids. [Connect](/connect/).

[mcp.json](/data/protocols/mcp.json) · [agent-client-protocol.json](/data/protocols/agent-client-protocol.json) · [Markdown](/compare/mcp-vs-agent-client-protocol/index.md)