reflection

You can turn experiences into actionable lessons by applying a structured reflection process that reviews actions, evaluates results, analyzes what worked, extracts key takeaways, and saves lessons to memory storage for future improvement.

The best way to trigger automated self-reflection is by using specific commands like /reflect, or relying on passive triggers such as session ends, pre-compaction events, or encountering errors and user feedback during a task.

Structured reflection integrates with memory storage by saving extracted lessons as categorized memory entries with defined importance levels, ensuring past error patterns and successful strategies are retrieved to prevent repeating mistakes.

You should perform a detailed reflection when encountering failures, errors, or reaching important milestones, whereas a quick mental review or short reflection suffices for general or complex tasks.

Yes, you can customize reflection depth based on task type, ranging from no reflection for simple queries to detailed reflection with memory recording for failures and significant project milestones.

To analyze failures and feedback effectively, ask what the root cause of the error is, how to avoid similar patterns, what the user truly wanted, where your understanding deviated, and how to better grasp user needs.