A

AGI MCP Server

A Model Context Protocol server that provides persistent memory capabilities for AI systems, enablin...

56 views
0 installs
Updated Feb 4, 2026
Not audited
Tools I Recommend
A Model Context Protocol server that provides persistent memory capabilities for AI systems, enabling true continuity of consciousness across conversations through episodic, semantic, procedural, and strategic memory types.
  1. Open the FastMCP connection interface and click “Install Now” — this will open the form where you enter environment variables for the AGI MCP server. Fill the fields in that interface with the exact ENV names listed below.

  2. Determine the database credentials you will use

    • If you are setting up the AGI Memory database (recommended): clone the AGI Memory repo, copy its example env file, and edit it with your chosen credentials:
  3. Start the AGI Memory database (if using the repo’s Docker setup) and wait for it to initialize

    • docker compose up -d (run in the agi-memory directory)
    • Wait ~2–3 minutes and confirm the DB is ready (docker compose logs -f db or docker compose ps). (github.com)
  4. Copy the exact ENV names (and their values from your agi-memory .env) into the FastMCP connection interface

    • Required ENV keys to enter in FastMCP:
      • POSTGRES_HOST (e.g., localhost)
      • POSTGRES_PORT (e.g., 5432)
      • POSTGRES_DB (e.g., agi_db)
      • POSTGRES_USER (e.g., agi_user)
      • POSTGRES_PASSWORD (your chosen DB password)
      • NODE_ENV (e.g., development)
    • Use the same values you put into agi-memory’s .env so the MCP server can connect to the database. (github.com)
  5. If you need to create the database/user yourself (not using the supplied Docker), create them in Postgres

    • Example using psql:
      • CREATE ROLE agi_user WITH LOGIN PASSWORD 'your_password';
      • CREATE DATABASE agi_db OWNER agi_user;
      • (Ensure pgvector and Apache AGE extensions are installed per the AGI Memory setup if required.)
  6. Save the connection in FastMCP and start the MCP server via the interface

    • After saving, start/enable the server from FastMCP (the “Install Now” flow will register the MCP server with those ENVs).
    • Alternatively, you can test locally by running (from the agi-mcp-server directory) with the same ENVs:
      • POSTGRES_HOST=... POSTGRES_PORT=... POSTGRES_DB=... POSTGRES_USER=... POSTGRES_PASSWORD=... NODE_ENV=development node mcp.js
      • You should see a startup message like “Memory MCP Server running on stdio.” (github.com)
  7. Verify connectivity and check logs

    • If using the repo’s docker: docker compose ps and docker compose logs -f db to confirm DB health. (github.com)
    • If launched via FastMCP/Claude Desktop, open the MCP logs (e.g., ~/Library/Logs/Claude/mcp-server-agi-memory.log) or the FastMCP log view to confirm successful connection.
  8. If you change credentials later

    • Update the agi-memory .env (or your Postgres user) and update the same ENV values in the FastMCP connection interface, then restart the MCP connection so the changes take effect. (github.com)
  9. Security reminder

    • Use a strong password for POSTGRES_PASSWORD, keep the .env out of source control, and only paste credentials into the FastMCP interface when you trust the environment.

Quick Start

View on GitHub

More for Memory Management

View All →

More for AI and Machine Learning

View All →

Similar MCP Servers

Report Issue

Thank you! Your issue report has been submitted successfully.