Dhee — The Cognition Layer That Turns Your Agent into a HyperAgent
Dhee is an open-source cognition layer for self-improving agents. It adds persistent memory,
performance tracking, insight synthesis, and prospective memory so agents can remember what happened,
learn from outcomes, and continue work across tools and sessions.
What Dhee adds
- Persistent memory: store facts, preferences, and observations with retrieval that survives rephrasing.
- Performance tracking: measure task outcomes, detect regressions, and surface warnings early.
- Insight synthesis: convert what worked and what failed into reusable learnings.
- Prospective memory: remember to do something later when the right trigger appears.
Interfaces
Dhee works through MCP, Python, CLI, and Docker, and integrates with Claude Code, Cursor, Codex, Claude Desktop, and custom runtimes.
View on GitHub