gsd-extract-learnings

Decision extraction pulls explicit choices and rationale from completed artifacts like PLAN.md and VERIFICATION.md, consolidating them into a single LEARNINGS.md file. It prevents valuable decisions and lessons from being lost after a phase completes.

Surprise detection in retrospective workflows surfaces unexpected outcomes and deviations from completed phase documentation. It consolidates these surprises into a LEARNINGS.md file, informing future work and preventing valuable deviations from being lost.

Consolidate end-of-phase wrap-ups by running the extract-learnings workflow with a phase-number argument. This processes planning through UAT artifacts and produces a structured LEARNINGS.md file at the workflow's configured location.

You need completed phase artifacts including PLAN.md, SUMMARY.md, VERIFICATION.md, UAT, and STATE documents. These provide the explicit choices, recurring issues, and unexpected outcomes necessary to generate a consolidated LEARNINGS.md file.

Workflow automation captures decision mining insights by executing the extract-learnings workflow on phase outputs. It identifies recurring patterns and rationale, consolidating them into a durable LEARNINGS.md artifact for team reuse.

Avoid automated learnings extraction when phase artifacts are incomplete or missing required PLAN, SUMMARY, VERIFICATION, UAT, and STATE documents. The workflow depends on these completed inputs to accurately mine decisions, patterns, and surprises.