What is MCP Memory Server?
The official MCP Memory Server gives Claude a persistent knowledge graph stored as a local JSON file. The AI can create entities, add observations to them, create relations between entities, and search the graph across conversation sessions.
Read the full overview
It is designed for workflows where the AI accumulates knowledge about a project, a codebase, or a domain over many conversations and needs to recall that context in future sessions without it being pasted into every prompt.
Persistent knowledge graph memory for Claude across conversations
Key capabilities
Entity and relation graph
Store named entities with type labels (person, project, concept) and typed relations between them. The graph is queryable by entity name, relation type, or full-text search across all observations.
Persistent observations
Add factual observations to any entity and retrieve them in later sessions. Observations are timestamped so the AI can distinguish stale facts from recent ones.
Where MCP Memory Server fits
Compatible with Claude Desktop, Cursor, Cline, and any MCP-compatible client. Install via npm (npx) or pip (uvx) depending on the server implementation language.
Positioning and limitations
The standard solution for adding long-term memory to Claude Desktop workflows. Works well for personal knowledge management, ongoing project tracking, and any use case where the AI should remember context that was established in a previous session.
MCP servers extend AI clients beyond their training data by giving them access to live systems: local files, databases, APIs, and services. Each server adds a set of typed tools the AI can call, inspect results from, and chain into multi-step workflows, all within a single conversation context.
“Persistent knowledge graph memory for Claude across conversations”
Technical details
- Access
- See vendor
- Source model
- Open source
- License
- Verify license
- Founded
- Not specified
- Pricing model
- Free
- API
- Not specified
- Website
- github.com ↗
- GitHub
- Source repository ↗
What to verify
Start from the users, systems and operating responsibilities the tool needs to support.
- Confirm current features, licensing and support terms.
- Validate deployment, data location and access control.
- Test integrations, exports and recovery workflows.
Install
npx -y @modelcontextprotocol/server-memoryReviews of MCP Memory Server
No published reviews yet.
Loading review form…