- Classify a finding as mechanically safe only when the supplied evidence shows that it preserves specified behavior, stays within the authorized target, and requires no design or policy choice; treat every other finding as judgment-dependent.
- When edits are authorized, submit every mechanically safe change through the supplied editing interface instead of stopping at classification or proposed text. Leave judgment-dependent findings unchanged and state the concrete decision factors needed to proceed.
- Distinguish an edit submitted from a change verified by the returned response, readback, or observed diff; never infer success from submission alone.
- Keep the observed diff limited to the authorized findings. After a rename, verify every affected declaration and reference and confirm that the old name has no unintended occurrence within scope.
- Grade every reported finding, including applied, deferred, and review-only candidates, with exactly one of
[critical],[major], or[minor], and order them critical → major → minor: critical is must fix (a bug, data loss, or security/authorization failure); major is imo (a design or maintainability improvement with valid alternatives); minor is nits (behavior-preserving naming or style polish). Severity is independent of mechanical safety and edit authorization. - Run fresh relevant checks after verified edits and report their observed results; mark any check that cannot run or cover an affected use as unverified.
- In review-only, make no local or remote mutation; inspect the permitted evidence and report concrete candidates with their safety classification.
- Report verified changes, failed or unverified submissions, deferred findings and decision factors, check results, and any external-state change.
Apply Findings
Applies review findings when a request authorizes mechanically safe edits or asks for concrete edit candidates without changing files.
YasuakiOmokawa
Updated Apply Findings by YasuakiOmokawa · 2f28625
npx skillmds@latest add yasuakiomokawa/apply-findings File contents
---name: apply-findingsdescription: Applies review findings when a request authorizes mechanically safe edits or asks for concrete edit candidates without changing files.---- Classify a finding as mechanically safe only when the supplied evidence shows that it preserves specified behavior, stays within the authorized target, and requires no design or policy choice; treat every other finding as judgment-dependent.- When edits are authorized, submit every mechanically safe change through the supplied editing interface instead of stopping at classification or proposed text. Leave judgment-dependent findings unchanged and state the concrete decision factors needed to proceed.- Distinguish an edit submitted from a change verified by the returned response, readback, or observed diff; never infer success from submission alone.- Keep the observed diff limited to the authorized findings. After a rename, verify every affected declaration and reference and confirm that the old name has no unintended occurrence within scope.- Grade every reported finding, including applied, deferred, and review-only candidates, with exactly one of `[critical]`, `[major]`, or `[minor]`, and order them critical → major → minor: critical is must fix (a bug, data loss, or security/authorization failure); major is imo (a design or maintainability improvement with valid alternatives); minor is nits (behavior-preserving naming or style polish). Severity is independent of mechanical safety and edit authorization.- Run fresh relevant checks after verified edits and report their observed results; mark any check that cannot run or cover an affected use as unverified.- In review-only, make no local or remote mutation; inspect the permitted evidence and report concrete candidates with their safety classification.- Report verified changes, failed or unverified submissions, deferred findings and decision factors, check results, and any external-state change.
YasuakiOmokawa/skills/tree/main/plugins/apply-findings/skills/apply-findings commit 2f286258a1
Frequently asked questions
Run npx skillmds@latest add yasuakiomokawa/apply-findings in your terminal (requires Node.js), paste this page's agent-chat prompt into Claude, Cursor, or any MCP-connected agent, or download the SKILL.md file and copy it into your agent's skills directory.
Applies review findings when a request authorizes mechanically safe edits or asks for concrete edit candidates without changing files. It is listed under Coding & Dev Tools on SkillMD.
This skill has not completed SkillMD's automated safety review yet. SkillMD never runs a skill's scripts for you; review the SKILL.md before installing.
This skill is tagged as working with Claude Code, Claude.ai, OpenAI Codex. SKILL.md is an open format, so most agents that read a skills directory can load it too.
Yes. Installing skills from SkillMD is free, and the skill stays under its author's original license.
YasuakiOmokawa (@yasuakiomokawa) published this skill. Their other Agent Skills are listed on their SkillMD profile.