kraklabs/mie
View on GitHub ↗Persistent memory graph for AI agents. Facts, decisions, entities, and relationships that survive across sessions, tools, and providers. MCP server — works with Claude, Cursor, ChatGPT, and any MCP client.
44 ★5 forksGoUpdated 5mo ago
What you need to know
Memory Intelligence Engine: a shared, persistent knowledge graph (CozoDB) that multiple AI agents read and write across sessions and providers.
Install
brew tap kraklabs/mie && brew install mie
Key features
- ✓12 MCP tools: store, bulk_store, query, conflicts, export, repair, status
- ✓Typed nodes: facts, decisions, entities, events, topics
- ✓Semantic search (optional Ollama embeddings), graph traversal, conflict detection
- ✓Agent-as-evaluator pattern - zero server-side inference cost
- ✓Server + daemon architecture with Unix domain socket
- ✓RocksDB/SQLite/mem storage engines
Best for
Teams who want cross-agent shared memory without extra inference costs.
Platforms: GoClients: Claude Code · Cursor
Reviewed 2026-08-11
Topics
ai-agentsai-memoryai-toolsdeveloper-toolsembeddingsgraph-databaseknowledge-graphllmmcpmcp-servermodel-context-protocolsemantic-search
- Stars
- 44★
- Forks
- 5
- Language
- Go
- License
- AGPL-3.0
- Created
- 2026-02-05
- Last push
- 2026-03-17