retrospective-agent

To extract reusable patterns from completed agentic tracks, analyze execution data like commits, task outcomes, and errors. This process identifies successful strategies and common pitfalls, updating knowledge bases with documented solutions for continuous workflow improvement.

Automated retrospective analysis improves workflow optimization by reviewing track metadata and specifications to log errors and identify friction. It generates skill enhancement proposals and updates knowledge bases, enabling future tasks to bypass documented pitfalls.

Generating a comprehensive development retrospective requires access to the track plan, metadata, commits, and specifications. This data allows the analysis to extract learnings, identify patterns, and log errors for continuous improvement.

You should run an automated learning extraction process after each track completion. Analyzing the finished execution data allows you to document successful strategies and update error logs, making future similar feature development smoother.

The best way to log errors for continuous improvement is to analyze completed tracks and record new error patterns into an errors.json file. This captures encountered errors and their fixes for reuse in future workflows.