capture-learning

To capture learnings for agent memory, summarize the current session insights, structure what worked and what didn't, and write them to agent-specific memories or project rules for future reference.

The best way to structure project retrospective insights is to organize what was learned into formats suitable for quick reference and auditing before writing them to memory logs.

Memory calibration maintains knowledge-base relevance by recalibrating stale items over time, ensuring that stored learnings and project rules remain accurate and continuously support ongoing sessions.

Yes, you can use session reflections to update project rules directly by structuring insights from what worked and what didn't, then storing them in project rules for ongoing calibration.

You need to recalibrate stale items in an agent memory store when previous learnings lose relevance, ensuring ongoing calibration maintains the accuracy of knowledge-base entries for future sessions.