lemmalog
View on GitHubA Datalog engine for LLM agent memory: stratified rules, provenance-tracked facts, incremental derivation, and an MCP server that lets your harness use it as a shared brain.
A Rust Datalog engine for persistent LLM-agent memory, with incremental rule evaluation, temporal and provenance-tracked facts, hybrid retrieval, and an MCP server for agent clients.
Use Cases
Persistent memory for LLM agentsDeriving and querying facts with Datalog rulesTracing conclusions back to source episodesMaintaining temporal facts and updatesSharing agent memory through MCPRetrieving relevant context for conversations
Built With
- Language
- Rust
- Frameworks
- MCP · OpenAI-compatible API
Tags
agent memory · Datalog · MCP · knowledge graphs · provenance · incremental reasoning · temporal reasoning · hybrid retrieval · embeddings · Rust