reflect-on-agent-sessions

To identify recurring issues in agent sessions, this Skill systematically analyzes autosearch-indexed session data to surface workflow problems and generates a structured markdown report with evidence and reasoning. It produces one report per issue family.

The best way to report automation reliability problems is to analyze autosearch-indexed sessions to surface recurring workflow patterns. This Skill generates a human-friendly markdown report containing evidence, reasoning, and representative incidents.

Yes, you need autosearch and access to session/message data to analyze agent sessions. The Skill refreshes the autosearch index across a specified analysis window to accurately identify recurring workflow problems and generate evidence-based reports.

The analysis generates evidence by scanning autosearch-indexed sessions within a defined window, counting times seen, and extracting representative incidents. It outputs a machine-readable markdown report detailing each issue family with structured reasoning.

Yes, you can produce multiple issue reports from a single analysis. The Skill operates across an analysis window and generates one distinct markdown report per identified issue family, ensuring each recurring problem is documented separately with evidence.

The session analysis report uses a structured markdown format for documenting agent incidents. This machine-readable and human-friendly output includes evidence, reasoning, and the frequency of recurring workflow problems to improve automation reliability.