Official MCP Registry · published 3h ago

Hivemind

io.github.monochrome-muncher/hivemind

Shared long-term memory for a team's AI agents. What one agent learns, the others can find.

Version: 2.18.1
Runs: Locally
Packages: oci
Published: Nov 8, 2026

Get started

Add it to your agent

claude mcp add hivemind -e HIVEMIND_DATABASE_URL=… -e HIVEMIND_EMBEDDING_ENDPOINT=… -- docker run -i --rm ghcr.io/monochrome-muncher/hivemind:v2.18.1

Run in your terminal. Add --scope project to share it with your team through .mcp.json.

codex mcp add hivemind --env HIVEMIND_DATABASE_URL=… --env HIVEMIND_EMBEDDING_ENDPOINT=… -- docker run -i --rm ghcr.io/monochrome-muncher/hivemind:v2.18.1

Saved to ~/.codex/config.toml.

{
  "mcpServers": {
    "hivemind": {
      "command": "docker",
      "args": [
        "run",
        "-i",
        "--rm",
        "ghcr.io/monochrome-muncher/hivemind:v2.18.1"
      ],
      "env": {
        "HIVEMIND_DATABASE_URL": "…",
        "HIVEMIND_EMBEDDING_ENDPOINT": "…"
      }
    }
  }
}

Add to ~/.cursor/mcp.json (or .cursor/mcp.json in a project).

{
  "servers": {
    "hivemind": {
      "type": "stdio",
      "command": "docker",
      "args": [
        "run",
        "-i",
        "--rm",
        "ghcr.io/monochrome-muncher/hivemind:v2.18.1"
      ],
      "env": {
        "HIVEMIND_DATABASE_URL": "…",
        "HIVEMIND_EMBEDDING_ENDPOINT": "…"
      }
    }
  }
}

Add to .vscode/mcp.json, then start it from the MCP view in Copilot Chat.

gemini mcp add hivemind docker run -i --rm ghcr.io/monochrome-muncher/hivemind:v2.18.1

Saved to .gemini/settings.json.

Before you start

Environment variables

HIVEMIND_DATABASE_URLrequiredsecretPostgres 16+ with pgvector 0.8+, for example postgresql://user:password@host:5432/hivemind. Migrations run when the container starts.
HIVEMIND_EMBEDDING_ENDPOINTrequiredAn OpenAI-compatible embeddings endpoint, for example a local vLLM at http://host.docker.internal:8001/v1.
HIVEMIND_EMBEDDING_MODELThe embedding model name the endpoint serves.
HIVEMIND_EMBEDDING_DIMVector width; must match the model and stay the same for the life of the database (at most 2000).
HIVEMIND_EMBEDDING_API_KEYsecretAPI key for the embedding endpoint, if it needs one.