Skip to main content

Serve ACP agents over MCP

Project description

ACP to MCP Adapter

Connect ACP Agents to MCP Applications Seamlessly

Apache 2.0 Follow on Bluesky Join our Discord LF AI & Data

The ACP to MCP Adapter is a lightweight standalone server that acts as a bridge between two AI ecosystems: Agent Communication Protocol (ACP) for agent-to-agent communication and Model Context Protocol (MCP) for connecting AI models to external tools. It allows MCP applications (like Claude Desktop) to discover and interact with ACP agents as resources.

Key Features

  • Makes ACP agents discoverable as MCP resources
  • Exposes ACP agent runs as MCP tools
  • Bridges two agent ecosystems with minimal configuration

Requirements

  • Python 3.11 or higher
  • Installed Python packages: acp-sdk, mcp
  • An ACP server running (Tip: Follow the ACP quickstart to start one easily)
  • An MCP client application (We use Claude Desktop in the quickstart)

Quickstart

1. Run the Adapter

Start the adapter and connect it to your ACP server:

uvx acp-mcp http://localhost:8000

[!NOTE] Replace http://localhost:8000 with your ACP server URL if different.

Prefer Docker?
docker run -i --rm ghcr.io/i-am-bee/acp-mcp http://host.docker.internal:8000

Tip: host.docker.internal allows Docker containers to reach services running on the host (adjust if needed for your setup).

2. Connect via Claude Desktop

To connect via Claude Desktop, follow these steps:

  1. Open the Claude menu on your computer and navigate to Settings (note: this is separate from the in-app Claude account settings).
  2. Navigate to Developer > Edit Config
  3. The config file will be created here:
  • macOS: ~/Library/Application Support/Claude/claude_desktop_config.json
  • Windows: %APPDATA%\Claude\claude_desktop_config.json
  1. Edit the file with the following:
{
  "mcpServers": {
    "acp-local": {
      "command": "uvx",
      "args": ["acp-mcp", "http://localhost:8000"]
    }
  }
}
Prefer Docker?
{
  "mcpServers": {
    "acp-docker": {
      "command": "docker",
      "args": [
        "run",
        "-i",
        "--rm",
        "ghcr.io/i-am-bee/acp-mcp",
        "http://host.docker.internal:8000"
      ]
    }
  }
}

3. Restart Claude Desktop and Invoke Your ACP Agent

After restarting, invoke your ACP agent with:

use "echo" agent with the "Good morning!" input

Accept the integration and observe the agent running.

Screenshot of Claude Desktop invoking the echo agent

MCP Resources

[!TIP] ACP agents are also registered as MCP resources in Claude Desktop.
To attach them manually, click the Resources icon (two plugs connecting) in the sidebar, labeled "Attach from MCP", then select an agent like acp://agents/echo.

How It Works

  1. The adapter connects to your ACP server.
  2. It automatically discovers all registered ACP agents.
  3. Each ACP agent is registered in MCP as a resource using the URI: acp://agents/{agent_name}
  4. The adapter provides a new MCP tool called run_agent, letting MCP apps easily invoke ACP agents.

Supported Transports

  • Currently supports Stdio transport

Developed by contributors to the BeeAI project, this initiative is part of the Linux Foundation AI & Data program. Its development follows open, collaborative, and community-driven practices.

Project details


Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Source Distribution

acp_mcp-0.2.0.tar.gz (162.0 kB view details)

Uploaded Source

Built Distribution

If you're not sure about the file name format, learn more about wheel file names.

acp_mcp-0.2.0-py3-none-any.whl (9.0 kB view details)

Uploaded Python 3

File details

Details for the file acp_mcp-0.2.0.tar.gz.

File metadata

  • Download URL: acp_mcp-0.2.0.tar.gz
  • Upload date:
  • Size: 162.0 kB
  • Tags: Source
  • Uploaded using Trusted Publishing? Yes
  • Uploaded via: twine/6.1.0 CPython/3.12.9

File hashes

Hashes for acp_mcp-0.2.0.tar.gz
Algorithm Hash digest
SHA256 98aba723f81df7b0fcbde68216393d56865341379ce362784e3c99a7662179c7
MD5 793379daadb1710781a5bd5b8de4151e
BLAKE2b-256 684b8f9c92a513ee3ce8d822f070ee49f75852e2ee9d7d78bcdb57d7f5b5763f

See more details on using hashes here.

Provenance

The following attestation bundles were made for acp_mcp-0.2.0.tar.gz:

Publisher: release.yml on i-am-bee/acp-mcp

Attestations: Values shown here reflect the state when the release was signed and may no longer be current.

File details

Details for the file acp_mcp-0.2.0-py3-none-any.whl.

File metadata

  • Download URL: acp_mcp-0.2.0-py3-none-any.whl
  • Upload date:
  • Size: 9.0 kB
  • Tags: Python 3
  • Uploaded using Trusted Publishing? Yes
  • Uploaded via: twine/6.1.0 CPython/3.12.9

File hashes

Hashes for acp_mcp-0.2.0-py3-none-any.whl
Algorithm Hash digest
SHA256 c86f85480cb442f79ad81bd7c5ced978ac70b414598f27c1a4f449c09aca0069
MD5 d9cd1ea3f3fbb96b12242512507b2956
BLAKE2b-256 34b1676537b8596fdc5a7ad5c1d33f5b7fef295fbeefeb3a0b52806e39d1c092

See more details on using hashes here.

Provenance

The following attestation bundles were made for acp_mcp-0.2.0-py3-none-any.whl:

Publisher: release.yml on i-am-bee/acp-mcp

Attestations: Values shown here reflect the state when the release was signed and may no longer be current.

Supported by

AWS Cloud computing and Security Sponsor Datadog Monitoring Depot Continuous Integration Fastly CDN Google Download Analytics Pingdom Monitoring Sentry Error logging StatusPage Status page