Skip to content
MCP server

Metis

Governed memory that gives AI agents reviewed expert know-how only where conditions match

Runs locallyStatus and history
Install
Pricing
Open source
Platforms
CLI, self-hosted
Runs
Locally
Version
0.1.6
Listing updated
1 day ago

About Metis

Metis is an open-source MCP server for knowledge and memory. It runs locally on your machine and is started by your AI client.

Metis captures what experts notice and how they respond, then offers it to agents as a fourth kind of memory alongside procedures, facts and past events. Each record, called a tacit fragment, keeps its source and conditions for use. When a recorded action departs from the written procedure, a capture agent asks the expert one short question and the expert confirms the account in their own words.

Fragments are promoted only after named human reviewers, and a condition-aware gate returns a fragment only where its context matches, for example a centrifugal pump under high load, and blocks it with a reason elsewhere. Every capture, confirmation, review and retrieval is logged on a hash-linked evidence chain through the CHAP coordinator, and each fragment carries one of seventeen categories of know-how.

A bundled manufacturing pump demo uses supplied observations and needs no model server, keeping records in a local project directory. The README is clear that consent workflows, access control and reviewer identity come from your application. It is published by Brightbeam and installs from PyPI.

Install

Run in a terminal

claude mcp add metis --env METIS_HOME='<METIS_HOME>' -- uvx metis-memory

Add to claude_desktop_config.json

{
  "mcpServers": {
    "metis": {
      "command": "uvx",
      "args": [
        "metis-memory"
      ],
      "env": {
        "METIS_HOME": "<METIS_HOME>"
      }
    }
  }
}

Add to ~/.cursor/mcp.json

{
  "mcpServers": {
    "metis": {
      "command": "uvx",
      "args": [
        "metis-memory"
      ],
      "env": {
        "METIS_HOME": "<METIS_HOME>"
      }
    }
  }
}

Add to .vscode/mcp.json

{
  "servers": {
    "metis": {
      "type": "stdio",
      "command": "uvx",
      "args": [
        "metis-memory"
      ],
      "env": {
        "METIS_HOME": "<METIS_HOME>"
      }
    }
  }
}

Add to ~/.codeium/windsurf/mcp_config.json

{
  "mcpServers": {
    "metis": {
      "command": "uvx",
      "args": [
        "metis-memory"
      ],
      "env": {
        "METIS_HOME": "<METIS_HOME>"
      }
    }
  }
}

Add to your client's MCP settings (most use this shape)

{
  "mcpServers": {
    "metis": {
      "command": "uvx",
      "args": [
        "metis-memory"
      ],
      "env": {
        "METIS_HOME": "<METIS_HOME>"
      }
    }
  }
}

Configuration

  • METIS_HOMEThe Metis project directory to serve. A directory with no workspace yet gets the pump demo, so there is governed memory to try at once.Required

What it does

Tacit fragments
Records what an expert noticed, how they responded and the context.
Human review
Promotes fragments only after named reviewers approve them.
Condition-aware gate
Returns a fragment only where its stated conditions hold.
Evidence chain
Logs captures, reviews and retrievals on a hash-linked record.
Seventeen know-how kinds
Classifies fragments by category from procedural to sensory.
Local demo
Ships a pump example that runs without a model server.

Good for

  • Preserve an operator's judgement before they retire
  • Give an agent guidance only in matching situations
  • Audit which expert knowledge an agent used
  • Prototype tacit-knowledge capture in a plant

Pricing

Open source

Open-source toolkit from Brightbeam

Prices change. Check metis.brightbeam.works before you buy.

Metis FAQ

What is a tacit fragment?

A reviewed record of what an expert noticed, how they responded and the circumstances.

Can I try it quickly?

Yes. Install metis-memory and run metis demo manufacturing-pump-vibration; no model server is needed.

What does my application have to provide?

Capture tools, consent workflows, access control, reviewer identity and storage policy.

Is it open source?

Yes, per the project, published on PyPI by Brightbeam.

Is Metis free?

Yes. Metis is open source, so you can use and self-host it for free.

How do I add Metis to Claude Code?

Run this command in your terminal: claude mcp add metis --env METIS_HOME='<METIS_HOME>' -- uvx metis-memory

What are the best alternatives to Metis?

The closest alternatives to Metis are Codebase Memory, XERJ, Claude Code Ultimate Guide, Nautilus Compass Memory and note1.

Alternatives to Metis

  1. Codebase Memory

    MCP server

    Local MCP server that builds a code knowledge graph for AI coding agents

    Runs locally
    • Open source
    • 48.7K GitHub stars
    • Knowledge and memory
  2. XERJ

    MCP server

    Local search engine for AI agents with cited answers over your code and docs

    Runs locally
    • Open source
    • 19.2K GitHub stars
    • Knowledge and memory
  3. MCP server for searching the Claude Code Ultimate Guide from any MCP client

    Runs locally
    • Open source
    • 6.5K GitHub stars
    • AI models and LLM tooling
  4. Local-first long-term memory layer for AI agents, served over MCP

    Not checked yet
    • 5 tools
    • Remote
    • Open source
    • 1.3K GitHub stars
    • Knowledge and memory
  5. note1

    MCP server

    MCP server giving AI assistants search and summaries across your note1 meetings

    • 4 tools
    • Remote
    • Free trial
    • 1 GitHub star
    • Productivity