About this project
Grok Mem is a persistent memory system designed for AI agents, including Claude Code, OpenClaw, OpenCode, and others. It captures agent activities during sessions, compresses the information using AI, and injects relevant context back into future sessions to ensure continuity of work and decision-making.
Key capabilities include:
- Persistent Memory: Maintains context across different sessions so agents remember previous decisions and progress.
- Progressive Disclosure: A token-efficient retrieval strategy that uses a three-layer workflow (search, timeline, and get_observations) to minimize token costs.
- Hybrid Search: Combines full-text search via SQLite (FTS5) with semantic search using a Chroma vector database.
- Integration Tools: Provides MCP (Model Context Protocol) tools for natural language queries of project history.
- Management Interface: Includes a local worker service with a web viewer UI for real-time memory streaming.
- Privacy Controls: Supports the use of private tags to exclude sensitive content from storage.
The system utilizes five lifecycle hooks (SessionStart, UserPromptSubmit, PostToolUse, Stop, and SessionEnd) to automate the capture and retrieval of memory without requiring manual user intervention.
Comments
0 Rating appears after 10 ratings
Sign in to join the discussion.