# knowledgeGraph MCP server

> Provides a long-term memory system for AI agents using a knowledge graph with semantic embeddings and automated memory decay and growth.

- Repository: https://github.com/ChenLiangChong/knowledgeGraph
- Page: https://hostedmcp.org/servers/chenliangchong/knowledgegraph
- Transport: —
- Stars: 6
- License: MIT

## Tools (12)

- `store_knowledge`
- `connect_knowledge`
- `update_knowledge`
- `forget_knowledge`
- `search_memory`
- `traverse_graph`
- `list_knowledge`
- `record_experience`
- `recall_experience`
- `maintain_graph`
- `crystallize_skill`
- `memory_stats`

## FAQ

### How does the system handle memory decay?

It uses a two-component decay model (fast and slow exponential) combined with FSRS stability to simulate human-like forgetting.

### Can I import my existing knowledge into the graph?

Yes, you can import markdown files as knowledge graph nodes using provided scripts.

### What happens when the AI starts a new session?

The system can automatically run maintenance tasks like repairing dangling edges, cleaning indexes, and calculating memory decay.