Bulletproof Memory
A Write-Ahead Log (WAL) protocol skill that prevents agent context loss through automatic state persistence triggered by user input. Maintains SESSION-STATE.md as persistent working memory across compactions, session restarts, and mid-conversation context shifts.
🚀 Bulletproof Memory stops your agent from losing critical information. Instead of hoping your agent remembers to save details, this skill automatically captures concrete facts the moment users provide them — before responding. It uses a Write-Ahead Log (WAL) protocol that triggers on user input, not unreliable agent memory, ensuring nothing important disappears during compaction, session restarts, or mid-conversation distractions.
💡 Perfect for multi-turn conversations, long-running tasks, and workflows where context matters. Whether your agent is managing project details, collecting user preferences, or tracking decisions, Bulletproof Memory keeps SESSION-STATE.md as persistent "RAM" that survives context limits and session resets. Your agent stays sharp across conversations.
✨ The key advantage: rules fire automatically on what users say, not on what agents remember. No more lost context, no more "where were we?" — just reliable, persistent working memory.
Requirements
No additional requirements.