Keyoku

Keyoku is a persistent memory system for AI agents that focuses on helping OpenClaw remember useful context across sessions. Instead of leaving memory as a loose collection of notes, it combines storage, semantic search, knowledge graph building, and a proactive heartbeat layer that can surface what needs attention without waiting for an explicit prompt. The organization positions it as a local-first setup: the OpenClaw plugin handles recall and capture during conversations, while a local memory engine manages deduplication, decay, graph relationships, and signal-driven analysis. The workflow described on the site is practical and opinionated: recall memory before replies, capture new facts after messages, and use heartbeat checks to identify deadlines, patterns, or other changes worth surfacing. That makes Keyoku a strong fit for people who want more than simple note storage or vector search. It is designed for operators who want an agent memory layer with structured retrieval, proactive monitoring, and data kept on their own machine rather than in a hosted memory service.

Use this tool if you:
- want OpenClaw to retain useful context between sessions instead of starting fresh each time
- need semantic search plus a more structured memory model with entities and relationships
- want automatic capture of facts and proactive heartbeat-driven reminders or signals
- prefer memory data to stay on your own machine instead of relying on a hosted service
- are building longer-running agent workflows where memory quality affects usefulness over time