Skip to content
RememRemem Docs

Quickstart

  1. Pull the image

    Remem’s core is a single Rust service: remem-server — the REST API, embedded storage engine, and MCP server (Streamable HTTP at /mcp), all in one process on port 4545.

    Terminal window
    docker pull rememorg/remem-community:server-latest

  2. Create a compose file and start Remem

    See Docker Compose deployment for the full docker-compose.yml (healthcheck, volumes). Then:

    Terminal window
    docker compose up -d

    The remem-server healthcheck has a 60s start_period — the embedding model loads on first boot, so give it a minute before expecting healthy.

  3. Verify it’s running

    Terminal window
    curl http://localhost:4545/api/v1/health

    Expected:

    {"status":"healthy","message":"..."}

  4. Store your first memory

    No Authorization header needed here — the compose file above sets REMEM_ALLOW_AUTH_DISABLED=true for local dev. See Docker Compose: Enabling authentication before running this beyond your own machine.

    Terminal window
    curl -X POST http://localhost:4545/api/v1/memories \
      -H "Content-Type: application/json" \
      -d '{
        "content": "User prefers concise bullet-point answers over long prose.",
        "memory_type": "long_term",
        "tags": ["preference", "communication"],
        "importance": 0.8
      }'

    Response:

    {
      "id": "mem_01j...",
      "content": "User prefers concise bullet-point answers over long prose.",
      "memory_type": "long_term",
      "importance": 0.8,
      "tags": ["preference", "communication"],
      "created_at": "2026-07-01T12:00:00Z"
    }

  5. Search memories

    Terminal window
    curl -X POST http://localhost:4545/api/v1/memories/search \
      -H "Content-Type: application/json" \
      -d '{"query": "how should I format my responses?", "search_type": "hybrid", "limit": 5}'

    The hybrid search type combines keyword and semantic (embedding) scoring for best results.

  6. Connect it to Claude Desktop (optional)

    Claude Desktop speaks MCP over stdio, so run the remem-mcp binary as a stdio bridge to remem-server. Add to ~/Library/Application Support/Claude/claude_desktop_config.json:

    {
      "mcpServers": {
        "remem": {
          "command": "docker",
          "args": ["run", "--rm", "-i", "--network", "remem-network", "rememorg/remem-community:server-latest", "remem-mcp", "--server-url", "http://remem-server:4545"]
        }
      }
    }

    Restart Claude Desktop. Claude can now call store_memory, search_memories, and find_related automatically.

Next steps

Section titled “Next steps”