Most AI memory tools just store and retrieve. Keyoku goes further: auto-recall, auto-capture, knowledge graph, and a heartbeat system that acts on what matters without being asked. Three autonomy levels: observe, suggest, or act. Powered by a local Go engine with semantic search, memory decay, and dedup. TypeScript SDK & OpenClaw plugin.

Framer — Launch websites with enterprise needs at startup speeds.

Launch websites with enterprise needs at startup speeds.

I've been building AI agents for the past year and kept hitting the same wall: agents forget everything between sessions, and even with memory, they just sit there waiting to be asked.

Think about what makes us autonomous — we recall, we think, we act. Keyoku brings all three to your AI agents.

🧠 Memory system — auto-recall, auto-capture, knowledge graph, semantic search, decay, and deduplication

⚡ Heartbeat engine — continuously scans memory, detects what needs attention, and acts on it.

🎛️ Three autonomy levels — observe, suggest, or act. You control how far it goes.

🔒 Runs locally — Go engine, your data stays on your machine. Plugs directly into OpenClaw.

🗺️ What's next: plug-and-play for any agent framework, cloud platform for teams, more language support.

Give it a try and star the repo if it's useful — I'd love feedback on the heartbeat system and autonomy levels especially.