distill
View on GitHubContext intelligence layer for LLM agents: persistent memory with write-time dedup, sensitivity tagging, conflict detection, and hierarchical decay. ~12ms. No LLM calls. MIT.
Go-based context and memory service for LLM agents. It stores and deduplicates memories across sessions, detects conflicts, and compresses aging context; it also offers an API, CLI, MCP server, and Pinecone or Qdrant integrations.
Use Cases
Persist agent knowledge across sessionsDeduplicate retrieved contextDetect conflicting memoriesCompress aging conversation contextExpose memory and deduplication to MCP-compatible assistantsBuild deterministic context bundles from reviewed files
Built With
- Language
- Go
- Frameworks
- MCP · Pinecone · Qdrant
Tags
agent memory · persistent memory · context engineering · deduplication · conflict detection · context compression · RAG · MCP · deterministic