brain-savelisted

# Brain Save — Knowledge Preservation ## When to use Trigger proactively (don't wait for the user to ask) when: - User corrected your approach or pointed out an error - Debugging completed and root cause was found - An architectural or design decision was made - A non-trivial bug or gotcha was discovered - A brainstorming session produced actionable decisions - Expert information was received from a third party - The user explicitly asks to remember something - **Session ending** or a large multi-step task just completed (see Step 0) ## Procedure ### Step 0: Retrospective (session end / large task completed) If triggered at session end or after a multi-step task — do a quick self-scan before Step 1: - **Dead ends / wasted effort?** Wrong path tried, unnecessary retries → save as `mistake` - **Workflow improvement?** Subagent strategy that worked, better tool usage → save as `pattern` - **Implicit user signals missed?** User rephrased, showed frustration, had to redirect → save as `user` - **Tool/API surprise?** Unexpected behavior worth remembering → save as `pattern` - **Insights worth keeping?** Brainstormed ideas, third-party suggestions, design alternatives, project ideas — anything useful for future work that isn't an error → save as `decision` / `pattern` / `wiki` / `research` (pick by content type) - **Routing gap?** Did I pick the wrong tool and get corrected, repeat a tool choice manually, or use a tool for a recurring intent that has no route? → if the trigger