01
What this skill does
Use this skill to resolve thermo reports or structural feedback. It accounts for findings, decomposition, coverage, and line-count evidence, checks behavior parity, and verifies cohesive ownership before fixing a threshold concern. It reconciles stale items individually, continues authorized independent fixes, and preserves the source report and unrelated Git state. Follow-up review is bounded to affected changes when justified. Invocation is explicit-only: a user must invoke `$receiving-thermo-review`; matching prompts do not activate it automatically.
02
When to use it
- 01Verifying harsh structural findings against the current diff, line counts, call sites, and ownership boundaries.
- 02Resolving 350-line concerns through responsibility ownership and dependency seams, alongside decomposition, recursive coverage, and candidate sweep items.
- 03Applying scoped behavior-preserving simplifications while checking source, guard, output, and extension-point parity.
03
How it works
- 01
Read the available report, requested outcome, approved scope, baseline, and current checkout.
- 02
Create a disposition ledger from findings, decomposition gaps, recursive coverage rows, line-count rows, sweep rows, and blind spots.
- 03
Reconcile stale or inconsistent evidence per item, blocking only dependent edits and retaining open coverage gaps.
- 04
Map responsibilities and dependencies for threshold items, then select a cohesive remedy within approved boundaries.
- 05
Present a concise behavior-parity and verification plan, then continue already-authorized work.
- 06
Prioritize proven blockers while resolving independent items and recording justified waivers or carry-forward decisions.
- 07
Recompute affected line counts and run focused plus required checks; report remaining findings without recursive review cycles.
04
What you get back
- 01A full disposition ledger for every consumed thermo report item.
- 02Scoped structural fixes or evidence-backed challenges, waivers, and carry-forward decisions.
- 03Updated line-count, coverage, and behavior-parity status for affected surfaces.
05
Important boundaries
- 01Do not blindly apply thermo review feedback.
- 02Do not start broad architecture refactors without approved scope.
- 03Do not treat a lower count, dense formatting, or a move into a catch-all module as resolution of a 350-line finding.
- 04Do not hide potential user-visible regressions behind structural cleanup.
- 05Preserve existing staged contents; perform Git actions only within the user's specific authorization for this task.
- 06Do not claim the thermo gate is resolved until every finding, decomposition gap, threshold item, and open coverage row is accounted for.
06