literature-memory -fastmcp

This MCP server provides a lightweight source management layer for papers, webpages, books, videos, and blogs, with structured notes and the ability to link sources to knowledge graph entities. It supports multiple identifiers per source (e.g., arXiv, DOI, URL, ISBN) and tracks status (unread, reading, completed, archived). You can add sources, attach notes, and create relationships between sources and memory graph entities, such as indicating which paper introduces or discusses a concept. The server is designed to work alongside the MCP Memory Server for persistent graph storage, enabling rich connections between literature and entities in your knowledge graph. Common workflows include creating a source entry with multiple identifiers, composing notes to capture key takeaways, and linking the source to relevant graph entities to support retrieval and reasoning within the memory graph.

Prerequisites:

• Python 3.8+ installed on your system
• Access to the MCP Memory Server for persistent graph storage
• SQLite (or another supported database as indicated by your deployment)

Step-by-step installation:

1. Clone the repository or download the server files.
2. Create and prepare your database (example uses SQLite):
   • sqlite3 sources.db < create_sources_db.sql
3. Install Python dependencies (adjust to your project’s requirements):
   • python -m pip install -r requirements.txt
4. Start the server using the provided entry script (adjust paths as needed):
   • python source-manager-server.py
5. If using environment-based configuration, set up a .env file or export variables before launching, for example:
   • export SQLITE_DB_PATH=/path/to/sources.db
   • export MEMORY_SERVER_URL=https://<memory-server-host>:<port>

Notes:

• Ensure the MCP Memory Server is reachable and properly configured before starting the literature-memory server.
• Depending on your deployment, you may containerize this service or run it directly on your host.

Tips and common considerations:

• Environment variables: SQLITE_DB_PATH is required to point to your local SQLite database. MEMORY_SERVER_URL is optional if you’re connecting to a remote memory graph.
• Schema readiness: Ensure your database schema matches what the server expects (sources, source_notes, source_entity_links).
• Identifier handling: You can store multiple identifiers per source (e.g., arXiv and semantic scholar IDs) to enable flexible lookup.
• Entity linking: Use explicit relation types (discusses, introduces, extends, evaluates, applies, critiques) to keep graph relationships meaningful.
• Memory graph integration: The server relies on the MCP Memory Server for persistent storage of knowledge graph connections; verify connectivity and permissions.
• Troubleshooting: If the server fails to start, check log output for missing dependencies, database permissions, or port conflicts. Verify that the SQLITE_DB_PATH is writable by the running process.