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 cniska/skills --skill skill-testgit clone --depth 1 https://github.com/cniska/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/cniska/skills/skill-test)<a href="https://agentmods.dev/skills/cniska/skills/skill-test"><img src="https://agentmods.dev/badge/skills/cniska/skills/skill-test/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/cniska/skills/skill-test"><img src="https://agentmods.dev/badge/skills/cniska/skills/skill-test.svg" alt="Reviewed on agentmods" width="80" 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 | $0.00025 | $0.00572 |
| Opus 5.5 | $0.00010 | $0.00229 |
| Sonnet 5.5 | $0.00005 | $0.00114 |
| Haiku 4.5 | $0.00003 | $0.00057 |
Grade A, and why
skill-test 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 1mo 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 — 32 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Skill test
A skill is an instruction to a model, so the only real test is running it. Reading a changed skill tells you it reads well; running it against real projects tells you whether it holds up. Dry-run the skill via subagents on real repos, collect their friction reports, and act only on findings that converge.
Diversity beats count: three repos that differ in shape (stack, release model, docs-heaviness, process discipline) expose more than ten that rhyme. A gap that only one shape can reveal — a release with no command, a project with no spec — never shows up on lookalikes.
Workflow
- Pick 2–3 unlike repos. Choose for difference in the dimension the skill touches — not convenience. If the change affects release handling, pick repos that release differently; if it affects doc structure, pick a docs-heavy and a docs-light one.
- Run in test-only mode. One subagent per repo, in parallel. Each agent reads the changed skill in full, applies it to its repo, and writes output where it can be inspected without committing — uncommitted in the working tree, or a scratch path. Never commit or push from a test run.
- Demand a friction report. Each agent reports: what it produced, the judgment calls the skill left it to make, where the skill's guidance was ambiguous or fought the repo's reality, and what it did differently than the skill seemed to expect. The report is the product; the output artifact is evidence.
- Act on convergence. The same friction from independent runs is a defect in the skill — fix it. Friction from a single run is usually a legitimate per-project judgment call — leave it; encoding a rule for it cuts against delegating judgment to the model.
- Fold fixes and re-verify. Apply the convergent fixes to the skill, then re-run the worst-affected repo if the fix changed behavior materially. Clean up test outputs, or hand them to the user if they turned out better than what the repos had.
See also
skill-writing— the change this dry-run gates
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.
- 1mo ago Changed 9886595efb78
- 1mo ago Changed · +4 lines 2d7230fdd80d
- 2mo ago First seen · 28 lines · 25 tokens per session scan A 2797be519bd8
skill-test is a skill published in the GitHub repository cniska/skills (6 stars, last pushed 1mo ago), licensed MIT. It adds 25 tokens to every session and 572 once invoked, about $0.0001 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 2026-08-31.
skill-test compared with skills from other repositories
| Mod | Type | Stars | Last push | Licence | Scan | Tokens | Works with |
|---|---|---|---|---|---|---|---|
| skill-testcniska/skills | Skill | 6 | 1mo ago | MIT | A | 25 | Claude Code, Codex |
| imagegenopenai/codex | Skill | 134k | today | Apache-2.0 | A | 113 | Codex |
| playwright-climicrosoft/playwright | Skill | 99k | today | Apache-2.0 | A | 19 | Claude Code |
| local-ai-agentsmicrosoft/ai-agents-for-beginners | Skill | 79k | today | MIT | A | 200 | Claude Code, Codex |
| systematic-debuggingobra/superpowers | Skill | 314k | 3d ago | MIT | A | 21 | Claude Code |
| next-cache-components-adoptionvercel/next.js | Skill | 143k | yesterday | MIT | A | 95 | Claude Code |
| next-partial-prefetching-adoptionvercel/next.js | Skill | 143k | yesterday | MIT | A | 103 | Claude Code |