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 citedy/skills --skill skill-quality-evalgit clone --depth 1 https://github.com/citedy/skillsWrote 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/citedy/skills/skill-quality-eval)<a href="https://agentmods.dev/skills/citedy/skills/skill-quality-eval"><img src="https://agentmods.dev/badge/skills/citedy/skills/skill-quality-eval.svg" alt="Measured on agentmods" height="20"></a>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 | $0.00061 | $0.00605 |
| Opus 5 | $0.00030 | $0.00302 |
| Sonnet 5 | $0.00012 | $0.00121 |
| Haiku 4.5 | $0.00006 | $0.00060 |
Grade A, and why
skill-quality-eval 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 7d 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 — 59 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Skill Quality Evaluator
Validates .claude/commands/*.md files for quality, triggering accuracy, and best practices from Phil Schmid's (Google DeepMind) eval harness methodology.
What it checks
- YAML frontmatter — every command must start with
---block containingdescription: - Description length — minimum 40 characters (short descriptions don't trigger well)
- Jargon detection — flags internal terms that confuse users (dogfood, --dangerously, etc.)
- Directive language — descriptions must contain action verbs (Run, Fix, Scan, Extract, etc.) because passive descriptions don't trigger Claude's skill routing
- Duplicate detection — similar commands must have distinct descriptions (first 30 chars normalized)
How to run
Execute the bundled script — it requires zero dependencies (pure Node.js):
node .claude/skills/skill-quality-eval/scripts/run-eval.js [commands-dir]
- Default
commands-dir:.claude/commands(relative to cwd) - Pass an absolute path to scan a different directory
The script outputs a structured report with PASS/FAIL per file, summary stats, and specific fix suggestions.
When a check fails
| Check | How to fix |
|---|---|
| Missing frontmatter | Add ---\ndescription: "..."\n--- at top of file |
| Description too short | Expand to 40+ chars — explain WHEN to use and WHAT it does |
| Jargon detected | Replace internal terms with user-facing language |
| No directive verb | Add action verbs: "Run...", "Fix...", "Scan...", "Use when..." |
| Duplicate description | Make the first 30 chars unique between similar commands |
Customization
Edit scripts/run-eval.js to add:
- Custom jargon patterns (line ~15,
JARGON_PATTERNSarray) - Additional directive verbs (line ~30,
DIRECTIVE_PATTERNSarray) - Differentiation pairs (line ~50,
DIFF_PAIRSarray)
Integration with CI
The script exits with code 1 if any check fails, making it suitable for pre-commit hooks or CI pipelines:
What ships with it
2 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.
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.
- 7d ago First seen · 59 lines · 61 tokens per session scan A 8c8002615078
skill-quality-eval is a skill published in the GitHub repository citedy/skills (2 stars, last pushed 3mo ago), licensed MIT. It adds 61 tokens to every session and 605 once invoked, about $0.0003 per session on Opus 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 2026-08-31.
Other skills, from other repositories
kit-audit
Audit guardrails-kit compliance of the current session — run the transcript marker scorecard. Use when asked to audit compliance, check which guardrail rules fired or were missed, or score the session.
kit-migrate
Install or upgrade the guardrails kit into a project with an existing CLAUDE.md by executing MIGRATE.md exactly. Use when asked to migrate a CLAUDE.md, install the kit, or retrofit guardrails into a project.
kit-rearm
Post-compaction/resume recovery for the guardrails kit — re-run the SESSION.md S1 sequence manually. Use after a compaction, /resume, or when session context feels lost or stale.
kit-reasoning
Reasoning scaffolds pack — worked-example traces, counterexample hunts, invariants, options enumeration. Use when writing algorithms, recursion, parsers, state machines, index math, or choosing between design approaches.
kit-verify
Force the guardrails VERIFY echo protocol — prove every done/fixed/works claim with quoted command output. Use when asked to verify claims, check completion evidence, prove it works, or when "done" was claimed without proof.
kit-angular
Angular guardrails pack — signals, change detection, @for track, RxJS teardown, standalone defaults, plus UI state/a11y discipline. Use when writing Angular components, templates, or services.