Skip to main content
Glama

MCP Memory Service

docker-compose.http.yml2.16 kB
version: '3.8' # Docker Compose configuration for HTTP/API mode # Usage: docker-compose -f docker-compose.http.yml up -d services: mcp-memory-service: build: context: ../.. dockerfile: tools/docker/Dockerfile ports: - "${HTTP_PORT:-8000}:8000" # Map to different port if needed volumes: # Single data directory for all storage - ./data:/app/data # Model cache (prevents re-downloading models on each restart) # Uncomment the following line to persist Hugging Face models # - ${HOME}/.cache/huggingface:/root/.cache/huggingface # Optional: mount local config # - ./config:/app/config:ro environment: # Mode selection - MCP_MODE=http # Storage configuration - MCP_MEMORY_STORAGE_BACKEND=sqlite_vec - MCP_MEMORY_SQLITE_PATH=/app/data/sqlite_vec.db - MCP_MEMORY_BACKUPS_PATH=/app/data/backups # HTTP configuration - MCP_HTTP_PORT=8000 - MCP_HTTP_HOST=0.0.0.0 - MCP_API_KEY=${MCP_API_KEY:-your-secure-api-key-here} # Optional: HTTPS configuration # - MCP_HTTPS_ENABLED=true # - MCP_HTTPS_PORT=8443 # - MCP_SSL_CERT_FILE=/app/certs/cert.pem # - MCP_SSL_KEY_FILE=/app/certs/key.pem # Performance tuning - LOG_LEVEL=${LOG_LEVEL:-INFO} - MAX_RESULTS_PER_QUERY=10 - SIMILARITY_THRESHOLD=0.7 # Python configuration - PYTHONUNBUFFERED=1 - PYTHONPATH=/app/src # Offline mode (uncomment if models are pre-cached and network is restricted) # - HF_HUB_OFFLINE=1 # - TRANSFORMERS_OFFLINE=1 # Use the unified entrypoint entrypoint: ["/usr/local/bin/docker-entrypoint-unified.sh"] restart: unless-stopped healthcheck: test: ["CMD", "curl", "-f", "http://localhost:8000/api/health"] interval: 30s timeout: 10s retries: 3 start_period: 40s # Resource limits (optional, adjust as needed) deploy: resources: limits: cpus: '2.0' memory: 2G reservations: cpus: '0.5' memory: 512M

MCP directory API

We provide all the information about MCP servers via our MCP API.

curl -X GET 'https://glama.ai/api/mcp/v1/servers/doobidoo/mcp-memory-service'

If you have feedback or need assistance with the MCP directory API, please join our Discord server