Hippocampus Memory Core
Use this skill when you need long-term memory that survives the current turn.
Use It For
- storing key decisions and facts that should outlive the current session
- searching past architecture, deployment, or debugging context
- creating snapshots before risky changes
- restoring memory after migration or environment reset
Preferred Flow
- Check that Hippocampus configuration is present.
- Store only high-signal facts, not raw transcript spam.
- Search memory before repeating costly investigation.
- Snapshot before major refactors or rollout changes.
- Restore only into the correct agent or workspace scope.
Guidance
- Prefer deterministic retrieval over ad hoc summaries.
- Keep memory namespaced by workspace and agent.
- Use metadata to mark project, topic, and task boundaries.
- Treat memory as durable infrastructure, not scratchpad overflow.
- If configuration is missing, send the user to
hippocampus-openclaw-onboarding
instead of asking for a raw API key first.
Related
hippocampus-openclaw-onboarding for first-time setup
hippocampus-subagent-memory for isolated child-agent memory
@hippocampus/openclaw-context-engine for native OpenClaw lifecycle integration