reez455g/kerangka/verify-before-deleting-dead-code
Claude Code and agent skill by reez455g/kerangka
- community
Checking that the SKILL.md file is still in the repository.
Use before deleting anything labeled 'dead code', 'unused binding', or 'zero callers' — including claims from a prior audit or subagent — to avoid breaking live functionality on a false premise.
Install
npx vskill@latest install reez455g/kerangka/verify-before-deleting-dead-codeSKILL.md
Why
Instructions (or prior audit reports, including your own subagent scouts) sometimes assert something is "dead" or "unused" based on a shallow or literal grep. Deleting on that assumption can break live functionality — the opposite of a cleanup. Naive greps miss:
- String-concatenated calls:
api('GET', 'foo.php?id=' + id + '&mode=1')won't match a literal-string search formode=1if the grep pattern assumed a quoted literal. - Attributes/keys consumed by a compiler or runtime elsewhere in the codebase (e.g. template-engine hint attributes read by a separate loader script), not by the file that defines them.
- Identifiers produced via object shorthand or spread (
{ a, b, ...rest }) rather than explicitkey: value, which simple regex key-extraction misses.
Procedure before deleting anything called "dead"
- Grep the consumer, not just the definer. If X is a template attribute/hint, grep the file that parses the template (the runtime/compiler), not just the file that declares X.
- Grep loosely, then confirm tightly. Search for the bare identifier/string across the whole relevant surface (not one exact literal pattern) to catch string-concatenated or shorthand usages.
- Cross-reference systematically for template bindings. Extract every
{{ identifier }}(or equivalent) used in a template, and extract every place identifiers are produced (object literal keys, shorthand properties, spreads, class fields) in the logic that renders it. Diff the two sets — don't trust a first-pass regex; re-check any "orphan" hit by hand, since spread/shorthand syntax produces false positives constantly. - Never trust a prior audit's "zero callers" claim at face value, even your own from earlier in the same session — re-verify with a fresh, broader search before acting on it, especially before an irreversible-feeling delete.
- If truly dead after verification, delete freely — the point isn't to avoid deleting, it's to avoid deleting live code on a false premise. Report what you verified and how.
Output when you push back
If a delete request is based on a false "dead" premise: do the safe part of the request, decline only the unverified part, and show the exact grep/cross-reference evidence for why it's live — not just an assertion.
Security
- Security scanPassed, 90 of 100Pattern checks on the skill's files.
- Model reviewPassed
Quality evaluation
- Task evaluationNot runNo evaluation yet. Run one in Skill Studio.
Source
- SKILL.md in the repositoryChecking.omp/skills/verify-before-deleting-dead-code/SKILL.md
Popularity
- 7-day trend score
- 5
Not refreshed yet.
Works with
Any agent vskill installs to: 56 targets in all. These 39 have a page on this site.
- Claude Code
- Cursor
- GitHub Copilot
- Windsurf
- Codex
- Gemini CLI
- Cline
- Amp
- Roo Code
- Goose
- Continue
- Kilo Code
- Junie
- Trae
- OpenHands
- Qwen Code
- AI Agent
- Tabnine Chat
- Sourcegraph Cody
- Amazon Q
- Firebase Genkit
- Sweep
- Devin
- Phind
- Mentat
- Claude Desktop
- Aider
- GPT Pilot
- Plandex
- bolt.new
- v0
- Replit Agent
- Anse
- TypingMind
- BetterChatGPT
- LibreChat
- Warp AI
- Pieces
- Fig AI
Embed badge
[](https://verified-skill.com/skills/reez455g/kerangka/verify-before-deleting-dead-code)