self-reflection

To capture actionable reflections when workflows cause friction, append a concise entry to a daily markdown file. This stores insights structured around what happened, the root cause, and what would help to improve future agent runs.

Logging agent session learning is best done by writing structured daily markdown files. This approach captures root causes and concrete improvements, ensuring consistent learning across sessions to enhance efficiency and outcomes over time.

Automated reflection file retention works by pruning older daily markdown logs beyond ten days on startup. This mechanism keeps storage lean while preserving recent workflow friction entries for ongoing agent improvement.

A structured self-reflection template should include what happened, the root cause, and what would help. This format enforces actionable insights to avoid repeating workflow issues and improves future agent performance.

Daily workflow reflection entries are stored in dedicated markdown files under the agent/shared/reflections directory. Each file is named with the date in YYYY-MM-DD format to organize session learning chronologically.

No external dependencies are needed to start logging workflow reflections. The system operates independently by writing structured insights to local markdown files, requiring no additional components or setup.