Context Keeper

What is Context Keeper?

Context Keeper is an LLM-driven intelligent context memory management system that gives AI assistants a persistent "memory." It solves the problem of having to re-explain project backgrounds in every conversation, prevents AIs from "forgetting" past discussions, and helps teams capture organizational knowledge.

How to use Context Keeper?

Visit the open-source repository at github.com/redleaves/context-keeper for source code, documentation, and configuration instructions.

Key features of Context Keeper

• Wide recall + fine ranking architecture (80%+ recall, 75%+ accuracy)
• Four-dimensional unified context model (Project, Topic, Conversation, Code)
• Full MCP protocol support (HTTP/WebSocket/SSE)
• Automatic short-term to long-term memory conversion with LLM importance evaluation
• User/workspace isolation and 10,000+ QPS enterprise performance
• IDE integrations for Cursor and VSCode

Use cases of Context Keeper

• Code review assistant that automatically links historical discussions
• New hire onboarding with quick access to architecture decisions
• Team knowledge base that continually accumulates technical experience

FAQ from Context Keeper

How does Context Keeper improve recall and accuracy compared to traditional RAG?

It uses a two-stage architecture: wide recall (multi-dimensional parallel retrieval of ~100 candidates) achieves 80%+ recall, and fine ranking (LLM analysis) improves accuracy to 75%+.

What transport protocols does Context Keeper support?

HTTP, WebSocket, and SSE.

Which IDEs does Context Keeper integrate with?

Cursor and VSCode, with automatic context tracking.

Is Context Keeper suitable for enterprise use?

Yes, it provides user/workspace isolation and can handle 10,000+ QPS.