From fable-discipline
Phase 3 of the milestone workflow — cross-perspective adversarial review of the milestone diff, triage, numbered fix batches, recorded outcome.
How this command is triggered — by the user, by Claude, or both
Slash command
/fable-discipline:milestone-review <milestone-name> [base-ref]The summary Claude sees in its command listing — used to decide when to auto-load this command
Run **Phase 3 (Adversarial Review)** of the milestone workflow for: **$ARGUMENTS** Follow the `milestone-workflow` skill's Phase 3 instructions (`skills/milestone-workflow/references/phase-review.md`) and its cross-cutting discipline. Before anything: read this repo's `CLAUDE.md` for the verify commands and deploy model. Review the milestone diff with a DIFFERENT perspective than wrote it (cross-AI / multi-lens), triage findings CONFIRMED / PARTIAL / FALSE-POSITIVE (verify each against live code before fixing — kill what you can't reproduce), fix in atomic numbered batches re-running the...
Run Phase 3 (Adversarial Review) of the milestone workflow for: $ARGUMENTS
Follow the milestone-workflow skill's Phase 3 instructions
(skills/milestone-workflow/references/phase-review.md) and its cross-cutting
discipline.
Before anything: read this repo's CLAUDE.md for the verify commands and deploy
model.
Review the milestone diff with a DIFFERENT perspective than wrote it (cross-AI / multi-lens), triage findings CONFIRMED / PARTIAL / FALSE-POSITIVE (verify each against live code before fixing — kill what you can't reproduce), fix in atomic numbered batches re-running the verify gate after each, and record the confirmed/partial counts + remaining open items in the project's state doc. Never push after a failed run until a clean rerun is confirmed.
npx claudepluginhub petrkindlmann/fable-discipline --plugin fable-discipline