Getting it into your agent
One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.
npx skills add melodic-software/claude-code-plugins --skill audit-dead-codegit clone --depth 1 https://github.com/melodic-software/claude-code-pluginsWrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/skills/melodic-software/claude-code-plugins/audit-dead-code)<a href="https://agentmods.dev/skills/melodic-software/claude-code-plugins/audit-dead-code"><img src="https://agentmods.dev/badge/skills/melodic-software/claude-code-plugins/audit-dead-code/github.svg" alt="Measured on agentmods" height="20"></a>Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.
<a href="https://agentmods.dev/skills/melodic-software/claude-code-plugins/audit-dead-code"><img src="https://agentmods.dev/badge/skills/melodic-software/claude-code-plugins/audit-dead-code.svg" alt="Reviewed on agentmods" width="80" height="20"></a>- NVIDIA SkillSpector pass
What it costs to keep this loaded
Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5.1 | $0.00099 | $0.04020 |
| Opus 5.5 | $0.00040 | $0.01608 |
| Sonnet 5.5 | $0.00020 | $0.00804 |
| Haiku 4.5 | $0.00010 | $0.00402 |
Grade A, and why
audit-dead-code scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 6d ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
How it starts
The opening of the file, as written. The whole thing — 244 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Purpose
Find code nothing has touched in a long time, which a rotated tidying lane and a diff-scoped simplification pass cannot see, and answer what here is no longer reachable, and how confident are we? Every candidate is adjudicated against the dynamic-usage patterns static analyzers miss, so the report is a list of decisions a human can act on, not an analyzer dump to re-verify.
The four lanes are not peers
Present them as four lanes of unequal strength, each labeled. A report that treats them as equals is wrong even when every finding in it is right.
| Lane | Covers | Measured character | How it fails |
|---|---|---|---|
| knip | TS/JS: unused files, exports, types, enum members. Not class members (Knip 6 rejects --include classMembers) |
60% precision / 100% recall on trap fixtures | Unrestored, it manufactures false positives (a failed config load produced 2 phantom "unused files"); its ERROR: line goes to stderr, which the JSON reporter discards |
| vulture | Python: unused function, class, method, variable, attribute, plus unreachable code | 16.7% precision / 100% recall; all five trap classes false-positived at 100% | Low precision by construction. Its output is a worklist, never a verdict |
| gopls | Go: unexported symbols only (gopls check -severity=hint), the lane's declared coverage |
Correct on every measured symbol; 2.1s | An unresolved module graph suppresses hints: false negatives, the opposite of knip. Never describe the two degradations with one phrase |
| grep | Shell and PowerShell functions; function/class/const/let/var declarations in JS/TS no package.json root owns (ts-unreferenced-symbol); unreferenced source files (unreferenced-file) |
Shell symbols: 4/4 true positives, 0 false over 546 .sh files (shellcheck found 0 of the 4). JS/TS symbols: 66.7% precision / 66.7% recall (as recorded 2026-09-29). unreferenced-file precision unmeasured |
Low recall: $, -, . are non-word characters, so an adjacent hit reads as a reference and saves a possibly dead symbol. A JS/TS export only an entry point or outside consumer calls is a false candidate; a name in a comment saves a dead one |
What ships with it
22 files beside SKILL.md in the same directory: the scripts, references and assets a skill reads on demand. Not counted in the per-session cost; read them before you install if any of them is executable.
- context/adjudication.md 7.7 KB
- context/lanes.md 15 KB
- evals/evals.json 12 KB
- evals/fixtures/dead-and-dynamic.go 602 B
- evals/fixtures/dead-and-dynamic.py 1022 B runs code
- evals/fixtures/dead-and-dynamic.sh 1.2 KB runs code
- evals/fixtures/dead-and-dynamic.ts 674 B runs code
- evals/fixtures/gopls-hints.txt 77 B
- evals/fixtures/knip-degraded.stderr.txt 142 B
- evals/fixtures/knip-report.json 676 B
- evals/fixtures/py-entry.py 352 B runs code
- evals/fixtures/scan-uncovered-java.txt 1.0 KB
- evals/fixtures/standalone.mjs 333 B runs code
- evals/fixtures/ts-entry.ts 591 B runs code
- evals/fixtures/ts-orphan.ts 279 B runs code
- evals/fixtures/ts-used.ts 234 B runs code
- evals/fixtures/vulture-parse-error-input.toml 449 B
- evals/fixtures/vulture-parse-error.txt 128 B
- evals/fixtures/vulture-report.txt 276 B
- scripts/dead-code-scan.sh 42 KB runs code
- scripts/dead-code-scan.test.sh 62 KB runs code
- scripts/lib/dead-code-shapes.sh 21 KB runs code
What this file has done since we first saw it
Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.
- 6d ago Changed · +3 lines cc2b84474de1
- 11d ago Changed · -16 lines · -18 tokens per session a80cc125a712
- 18d ago Changed · -10 tokens per session 2dd1d3489499
- 19d ago Changed · +32 lines · +10 tokens per session dfd2ae001793
- 20d ago Changed · +29 lines · -59 tokens per session eb5f6db65df0
- 21d ago Changed · -78 tokens per session fd2311af442a
- 26d ago Changed 5c7cc3da296f
- 1mo ago Changed · -1 lines a7b5a6414e03
- 1mo ago Changed · +3 lines 799e1f86bfef
- 1mo ago First seen · 194 lines · 254 tokens per session scan A 5f2ea35343db
audit-dead-code is a skill published in the GitHub repository melodic-software/claude-code-plugins (26 stars, last pushed today), licensed MIT. It adds 99 tokens to every session and 4,020 once invoked, about $0.0004 per session on Opus 5.5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.
Other skills, from other repositories
adversarial-analysis
Analyze code to identify explicit contracts, implicit usage patterns, and realistic boundary conditions. Contains concrete formulas for calculating input realism limits. Use before generating adversarial tests.
benchmark-before-optimize
For performance work: measure before changing, profile to find bottlenecks, compare before and after.
handle-edge-cases
For boundary conditions: empty collections, zero values, recursive base cases, null checks, prevent crashes at edges.
validate-before-process
For input validation: check format and constraints before processing, fail fast with clear errors, defensive parsing.
display-grid-state
For 2D debugging: visualize grid/board state, show puzzle progress, make algorithm behavior visible.
claim-check
Claim checker that audits a coding agent's statements that tests pass or a build is clean against its own session transcripts: whether a matching run happened before the claim, whether it passed, and whether code changed after it. Use when the user asks whether the agent really ran the tests, how often it said tests…