Skip to main content

MCP server for Agent Brain — cross-source personal memory for any MCP-capable AI client (Claude Code, Claude Desktop, Cursor, Windsurf)

Project description

agentbrain-mcp

MCP server that gives any MCP-capable AI client (Claude Code, Claude Desktop, Cursor, Windsurf) access to your personal Agent Brain — cross-source memory that persists across tools, conversations, and devices.

Status: status.agentbrain.ch · Changelog: CHANGELOG.md · License: MIT

What it does

Three tools, registered as native MCP tools:

  • brain_recall(query, limit) — search your Brain for relevant memories
  • brain_store(content, memory_type, source_trust) — persist a memory
  • brain_status() — health check

The server is single-user: each install is bound to one workspace via env vars. If you have multiple brains (work/personal), run multiple instances.

Install

pip install agentbrain-mcp

Or from source:

git clone https://github.com/AgentBrainHQ/agentbrain-mcp
cd agentbrain-mcp
pip install -e .

Configure

Get your API key and workspace id from https://agentbrain.ch/settings. Export them:

export BRAIN_API_KEY='brain_xxxxxxxxxxxx'
export BRAIN_WORKSPACE_ID='xxxxxxxx-xxxx-xxxx-xxxx-xxxxxxxxxxxx'

Optional:

Env var Default Purpose
BRAIN_URL https://api.agentbrain.ch Brain API base URL (override for self-hosted)
BRAIN_TIMEOUT_S 45 Request timeout in seconds
BRAIN_DEFAULT_LIMIT 10 Default limit for brain_recall when not provided

Use with Claude Code

claude mcp add agentbrain --scope user \
  -e BRAIN_API_KEY=$BRAIN_API_KEY \
  -e BRAIN_WORKSPACE_ID=$BRAIN_WORKSPACE_ID \
  -- agentbrain-mcp

Then in any Claude Code session:

> brain_recall("previous decisions on auth")

Use with Cursor / Windsurf

Add to your MCP config (~/.cursor/mcp.json or equivalent):

{
  "mcpServers": {
    "agentbrain": {
      "command": "agentbrain-mcp",
      "env": {
        "BRAIN_API_KEY": "brain_xxxxxxxxxxxx",
        "BRAIN_WORKSPACE_ID": "xxxxxxxx-xxxx-xxxx-xxxx-xxxxxxxxxxxx"
      }
    }
  }
}

Use with Claude Desktop

Same JSON as Cursor, in ~/Library/Application Support/Claude/claude_desktop_config.json (macOS) or %APPDATA%\Claude\claude_desktop_config.json (Windows).

Test it runs

# Should print a config error if env not set — no network required
agentbrain-mcp

With env set it will wait for an MCP client to connect on stdio.

Run over HTTP / SSE

For Claude Desktop Remote Connectors or a self-hosted endpoint:

# Streamable HTTP (recommended for new deployments)
agentbrain-mcp --http

# Legacy SSE
agentbrain-mcp --sse

Single-tenant: the endpoint is bound to the API key and workspace id of the process that started it. Host behind your own reverse proxy (Traefik, Caddy, nginx) for TLS. Multi-tenant hosting (one endpoint, many users with different keys) requires a separate server and is not in scope here.

Development

pip install -e ".[dev]"
pytest

Build the package:

pip install build twine
python -m build
twine check dist/*

License

MIT — see LICENSE.

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

agentbrain_mcp-0.1.0.tar.gz (7.8 kB view details)

Uploaded Source

Built Distribution

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

agentbrain_mcp-0.1.0-py3-none-any.whl (9.0 kB view details)

Uploaded Python 3

File details

Details for the file agentbrain_mcp-0.1.0.tar.gz.

File metadata

  • Download URL: agentbrain_mcp-0.1.0.tar.gz
  • Upload date:
  • Size: 7.8 kB
  • Tags: Source
  • Uploaded using Trusted Publishing? No
  • Uploaded via: twine/6.2.0 CPython/3.12.13

File hashes

Hashes for agentbrain_mcp-0.1.0.tar.gz
Algorithm Hash digest
SHA256 91136b9c9da3d7d047673efc7a1553974f8d91f98b6c8d929ef1f03488fd4802
MD5 9a728bfe5011416811019e9571811900
BLAKE2b-256 2735b939034c3b6ead4556001497a8be2b5b2b91d3f1e63eaab989d0370daaed

See more details on using hashes here.

File details

Details for the file agentbrain_mcp-0.1.0-py3-none-any.whl.

File metadata

  • Download URL: agentbrain_mcp-0.1.0-py3-none-any.whl
  • Upload date:
  • Size: 9.0 kB
  • Tags: Python 3
  • Uploaded using Trusted Publishing? No
  • Uploaded via: twine/6.2.0 CPython/3.12.13

File hashes

Hashes for agentbrain_mcp-0.1.0-py3-none-any.whl
Algorithm Hash digest
SHA256 03d4ba492247c786ac8fb13d2b513ec5ba86a6cb86f137534035e08ac6c6a9ac
MD5 62644559c9a4905f319730ef1d30950b
BLAKE2b-256 172e146b953856d7fab637cdbf4830f41fadc427cf6cc04023d0f2d7d37bc619

See more details on using hashes here.

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