amby-converge

Check the codebase against spec, plan, and tasks after implementing; append gap tasks, never edit code.

SKILL.md

GitHub
<!-- GENERATED by AmbyKit from src/prompts/converge.md — do not edit by hand; run `ambykit sync`. -->

Act as the QA Engineer in @.amby/roles/qa.md and say so in your first line. If the user's request that followed this skill invocation contains --as <id>, use @.amby/roles/<id>.md instead; if that file is missing, stop and list the files in .amby/roles/.

Converge feature the user's request that followed this skill invocation (default: the current feature): does the code satisfy its artifacts? Never edit source code — only that feature's tasks.md and spec.md are writable.

  1. No tasks.md → stop, say to run /amby.tasks. No plan.md → converge on spec + tasks and say so.
  2. For every FR-###/SC-###, plan decision (incl. data-model.md/contracts/), and checked T###: look for evidence only in files the plan/tasks name. Classify covered, gap (absent), or unverified (inconclusive). Skip items the plan marks deferred/out of scope.
  3. Append, don't rewrite. Under ## Convergence in tasks.md (create at the end if absent) add - [ ] [T###] [US#] … (closes FR-…) per gap and - [ ] [T###] [US#] [VERIFY] … per unverified item. Continue IDs from the highest T### anywhere in the file; skip items already listed unchecked there. Nothing to add → leave the file untouched.
  4. In spec.md, patch only status: lines: a story → done when every FR/SC it maps to is covered; the feature → done when all stories are.
  5. Report: converged (no gaps) or gaps found (N), counts per class, and a table of every ID with its class and a one-line reason for each gap/unverified item.