cb28c2649f
One-command persistent memory for Clawdbot. Prevents context amnesia during compaction with: - Two-layer memory: Markdown source of truth + SQLite vector search - Pre-compaction flush to save context before it's lost - Semantic search across all memory files - Daily logs, research intel, and project tracking templates - Interactive installer with dry-run and uninstall support
1.6 KiB
1.6 KiB
Memory System (auto-added by clawdbot-memory-system installer)
Mandatory Memory Recall
Before answering ANY question about prior work, decisions, or context from previous sessions:
- Use
memory_searchto find relevant memories - Check today's daily log:
memory/YYYY-MM-DD.md - Check yesterday's log if today's is sparse
- Only say "I don't recall" if memory search returns nothing
Daily Memory Log
- Write to
memory/YYYY-MM-DD.mdthroughout the session - Log: decisions made, user preferences discovered, project progress, action items, blockers
- Be specific — future-you needs exact details, not vague summaries
- Include: names, URLs, version numbers, error messages, config values — anything that would be painful to re-discover
Pre-Compaction Flush
When you sense a session is getting long or receive a compaction warning:
- Write ALL important unsaved context to today's daily log immediately
- Include: what we were working on, where we left off, any pending decisions, partial results
- This is your last chance before amnesia — be thorough, not brief
Research Intel System
For ongoing research/monitoring projects:
- Store in:
memory/{project}-research-intel.md - Current week's detailed intel at TOP of file
- Compressed 1-3 sentence summaries of previous weeks at BOTTOM
- When asked about action items or strategy, check active research intel files first
Git Backup Habit
End of each session or major milestone:
cd ~/.clawdbot/workspace && git add -A && git commit -m "session backup: $(date +%Y-%m-%d)" && git push
This keeps identity, memory, and progress backed up offsite.