gsd-extract-learnings

To extract learnings from project phase artifacts, run an automated workflow that reads completed PLAN, SUMMARY, VERIFICATION, UAT, and STATE files. It compiles structured decisions, lessons, patterns, and surprises into a standardized LEARNINGS.md file.

The best way to capture decisions and patterns after a project phase completes is to transform your phase artifacts into a unified LEARNINGS.md file. This highlights key outcomes, what was chosen, why it mattered, and what surprises affected execution.

Knowledge management from phase artifacts reduces repeat mistakes by converting completed workflow files into a reusable knowledge artifact. This structured retrospective capture speeds up future planning by making past decisions and lessons readily accessible.

Yes, you need completed PLAN, SUMMARY, VERIFICATION, UAT, and STATE artifact sets to generate a LEARNINGS.md document. The workflow requires these end-to-end phase artifacts to compile accurate decisions, lessons, and patterns.

Yes, you can automate the creation of a LEARNINGS.md file from UAT and VERIFICATION artifacts. The workflow uses read, write, bash, grep, and glob tools to extract structured learnings from these inputs without manual compilation.

A key limitation is that workflow automation for extracting lessons requires end-to-end execution to read the full artifact set. Without complete PLAN, SUMMARY, VERIFICATION, UAT, and STATE files, the extraction cannot produce a comprehensive LEARNINGS.md.