.cursor/skills/superpowers/superpowers-subagents-final-check/SKILL.md
Closes the loop after subagent-driven code-quality or spec reviews by reconciling all reviewer feedback with the repo, then creating VOM tickets for valuable work that is still open. Use after a Task/code-reviewer chain, when finishing a multi-task plan, or when the user asks to process review feedback into tickets.
npx skillsauth add denerose/VeganMealPlanner superpowers-subagents-final-checkInstall this skill globally with one command. Works with Claude Code, Cursor, and Windsurf.
3 of 9 scanners reported clean
Some scanners were skipped, did not run, or reported a non-clean status. Review each row below.
Run this after implementer + spec + code-quality subagent chains (or any session that produced structured review feedback). Goal: nothing important is lost—either it is already fixed in the tree, or it is tracked on VOM.
vom from the project root (AGENTS.md, Bun tooling).From each review, extract rows with:
Merge duplicates across reviews of the same commit or feature.
For every row, decide:
| State | Action |
|-------|--------|
| Already implemented in current tree | Mark resolved; no ticket. |
| Partially addressed (e.g. comment added, not full fix) | Ticket only if meaningful work remains. |
| Operational / docs only | Prefer vom new -t docs when no code change is required yet. |
| Refactor / test stability | vom new -t refactor or -t task as appropriate. |
| Product or schema decision deferred | Ticket with clear “consider when …” scope, or skip with one-line note in the plan’s review-debt section. |
Do not file tickets for noise: pure praise, “expected” trade-offs (e.g. native binary for Argon2), or duplicates of existing tickets—grep .vom / recent vom output if unsure.
Align with Capturing code quality review feedback in .cursor/skills/superpowers/subagent-driven-development/SKILL.md:
Valid vom new types here: feature, bug, task, refactor, docs (not chore).
Example:
vom new -t task "Short imperative title" "1–3 sentences: context, constraint, done-when."
Follow .cursor/skills/vom-new/SKILL.md for title/description quality.
If a suggestion is under about two minutes of work (rename a misleading test, typo in comment), fix in the repo and do not open a ticket unless the user wants audit trail.
Ad-hoc docs under tests/ or src/: If a review or reconciliation finds a new README.md under tests/** or src/** that the user did not ask for, move the content to TESTING.md (or the appropriate existing doc) and delete the stray file—see .cursor/rules/documentation-no-ad-hoc-readmes-tests-src.mdc.
Give a short table:
Optionally append a Review debt subsection to the relevant plan under docs/superpowers/plans/ only if the user or plan already uses that pattern.
bun run check-all / ./scripts/check-all.sh runs integration tests too—use only when integration tests changed or unit tests are insufficient; default remains bun run check (see AGENTS.md).prisma/migrations/ to “fix” schema feedback—use schema + Prisma workflow per project rules.testing
Guide for continuing work on a VOM ticket - reminds agents to claim and check guidance before starting.
data-ai
Move VOM tickets with state done or deleted into tickets/done or tickets/deleted to keep the main list tidy.
testing
Review recent work for friction and pain points. Create or update tickets for issues that made the agent's job harder. Focus on project-specific improvements only.
tools
Reviews, amends, and improves a VOM implementation plan for a ticket in plan-needs-review using only the vom CLI for claims, guidance, comments, and state transitions; edits plan markdown under plans/ only. Use when the user wants a plan approved (moved to plan-approved), wants plan-needs-review handled end-to-end, or invokes plan review with extra notes or constraints in the prompt.