Memory

What is Memory?

Memory is an MCP server that provides persistent memory using a local knowledge graph. It lets AI assistants like Claude remember information about the user across chats by storing entities, relations, and observations.

How to use Memory?

Configure Memory in your MCP client (Claude Desktop or VS Code) using Docker or npx. Optionally set the MEMORY_FILE_PATH environment variable to customize the storage file. Use the provided tools (e.g., create_entities, create_relations, add_observations) and the knowledge-graph resource to read and modify the graph.

Key features of Memory

• Persistent memory via local knowledge graph
• Entities with types and atomic observations
• Directed relations in active voice
• CRUD tools for graph manipulation
• Readable knowledge graph resource with live updates
• Configurable JSONL file storage

Use cases of Memory

• Remembering user identity and preferences across sessions
• Storing recurring behaviors, habits, and interests
• Tracking personal and professional goals
• Maintaining relationship networks for context
• Customizing assistant responses with past information

FAQ from Memory

How does Memory persist data?

Data is saved to a local JSONL file. The default path is memory.jsonl in the server directory, but can be overridden with the MEMORY_FILE_PATH environment variable.

What tools does Memory provide?

Memory offers create_entities, create_relations, add_observations, delete_entities, delete_observations, delete_relations, read_graph, search_nodes, and open_nodes. All tools are designed for a knowledge graph.

What runtime is needed?

Memory can be run via Docker or using Node.js with npx. No external database is required.

How do I configure the storage file path?

Set the MEMORY_FILE_PATH environment variable in the MCP server configuration (e.g., in claude_desktop_config.json). The default location is memory.jsonl in the server directory.

Does Memory support VS Code?

Yes. Memory can be installed in VS Code using the provided one‑click buttons or by adding the configuration to the user or workspace MCP settings file.