Session History

Cleanup Review

python scripts/_cleanup.py is an optional cleanup helper for the Session History skill. It may recursively remove /tmp/_skill_run/*, /tmp/_skill_run/.cache. Show the exact paths first, explain whether they are skill-private temporary paths or user state, and run the cleanup only after the user confirms.

Session History

Search through past OpenClaw session transcripts (JSONL files in ~/.openclaw/agents/*/sessions/).

Quick Reference

# Search for conversations about a topic
python3 scripts/search_sessions.py "gclid pipeline error"

# List recent sessions
python3 scripts/search_sessions.py --list --days 3

# Search specific agent's history
python3 scripts/search_sessions.py "flight monitor" --agent main

# Wider time range
python3 scripts/search_sessions.py "quantum encryption" --days 30 --max-results 5

Workflow

1. Run search_sessions.py with the user's query terms to find relevant sessions
2. Use sessions_history tool with the sessionKey to pull full context from a match
3. If sessions_history doesn't work (old/closed sessions), read the JSONL file directly with read
4. Summarize what was found — don't dump raw transcripts

When to Use

• User says "remember when we discussed X?" or "we talked about Y last week"
• Resuming a thread that isn't captured in memory files
• Finding a decision, code snippet, or error from a past session
• Cross-referencing what was said vs what's in MEMORY.md

Tips

• Also check memory_search first — it indexes session transcripts too
• Combine both: memory_search for semantic matching, search_sessions.py for keyword/exact matching
• The script searches user AND assistant messages
• JSONL path format: ~/.openclaw/agents/{agent_id}/sessions/{session_uuid}.jsonl