extract-learnings

You can extract learnings by scanning decision logs, development sessions, and git commits to classify patterns, anti-patterns, and tooling pitfalls into a searchable knowledge base file.

Automating a retrospective knowledge base involves scanning raw materials like ADRs and decision logs, then classifying them into patterns and pain points to write a dated extract report.

Writing decision logs to a context store prevents repeating mistakes by capturing classified patterns and tooling pitfalls from recent sessions into a centralized, searchable learnings file.

Yes, you can generate optional PR suggestions to promote identified patterns into rules after extracting and classifying reusable knowledge from git histories and development sessions.

You need a .context/learnings directory to store the generated dated extract files, which organizes your knowledge base by naming reports with the specific date range.

Extracting knowledge from raw development sessions requires available raw materials like decision logs and git commits, limiting output to classified patterns and pain points written into a markdown file.