memory-setup-openclaw_skill

Overview

This skill configures and validates durable memory recall for OpenClaw agents so they can reliably reference past decisions, preferences, and todos. It walks through workspace layout, enabling memory search, testing recall, and troubleshooting to ensure persistent context works as intended.

How this skill works

It ensures long-term context is indexed by OpenClaw by placing curated notes in MEMORY.md and daily notes under memory/*.md, then enabling memory search in agents.defaults.memorySearch. After restarting the gateway, the skill validates that the agent performs memory_search followed by memory_get when needed and cites source paths or lines for auditability.

When to use it

• Enabling memory_search or memory_get for an agent workspace
• Setting up a MEMORY.md + memory/*.md workflow for persistent context
• Fixing poor or inconsistent memory recall from an OpenClaw agent
• Switching memory providers (local vs remote) or changing search thresholds
• Verifying memory tool policy and CLI memory status after config changes

Best practices

• Keep a curated MEMORY.md for long-term facts and use memory/YYYY-MM-DD.md for daily logs
• Place optional subfolders like memory/projects, memory/system, memory/groups to organize notes
• Enable memorySearch under agents.defaults.memorySearch, not at the top level
• Restart the gateway after config edits and confirm with openclaw status or openclaw memory status
• Prefer memory_search first, memory_get second (fetch only needed lines) and report that memory was checked

Example use cases

• Enable recall of user preferences and past decisions before answering follow-up questions
• Debugging low-quality memory matches by tuning minScore and maxResults
• Switching from a local memory provider to a remote provider and setting the API key
• Setting up an auditable workflow where agents cite MEMORY.md or daily note sources
• Validating that an agent runs memory_search and memory_get in a typical conversational flow