Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Reviews SKILL.md files against a 10-dimension quality rubric inspired by the Agent Skills specification. Use when auditing skills before publish or users ays review skill, score SKILL.md, skill quality audit. Triggers: reviewskill , skill rubric, audit SKILL.md.
.claude/skills/stijnman-skill-rubric-reviewer/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-04 | ✗→✓ | ▲ Improved | 2% | 0% |
| case-05 | ✗→✓ | ▲ Improved | -24% | 0% |
| case-17 | ✗→✓ | ▲ Improved | -51% | 0% |
| case-10 | ✗→✓ | ▲ Improved | -28% | 0% |
| case-13 | ✗→✓ | ▲ Improved | -37% | 0% |
Run before publishing any skill:
bashpython3 scripts/check_no_private_data.py python3 scripts/publish_safety_check.py python3 scripts/validate_collection.py
validate_collection.py completes successfullyhitl-approver referenced for destructive opsskill-evolverhyper-skill-testerprivacy-redactor| Failure | Response | |---------|----------| | Missing SKILL.md | Report path; do not score directory without file. |
This skill is designed for public distribution. Constraints:
Input: User request matching triggers above. Output: Structured result per workflow; local artifacts only unless user opts in.
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→fail | 6,329 | 10,452 | +65% | 1 | 1 | 0% | 305 | 1,046 | +243% | 0 | 0 | — |
case-02 | fail→fail | 6,656 | 2,972 | -55% | 1 | 1 | 0% | 1,124 | 877 | -22% | 0 | 0 | — |
case-03 | fail→fail | 5,922 | 3,696 | -38% | 1 | 1 | 0% | 272 | 943 | +247% | 0 | 0 | — |
case-04 | fail→pass | 13,255 | 6,935 | -48% | 1 | 1 | 0% | 1,774 | 1,816 | +2% | 0 | 0 | — |
case-05 | fail→pass | 14,808 | 7,219 | -51% | 1 | 1 | 0% | 2,349 | 1,784 | -24% | 0 | 0 | — |
case-06 | fail→fail | 8,240 | 4,189 | -49% | 1 | 1 | 0% | 1,322 | 1,216 | -8% | 0 | 0 | — |
case-07 | pass→pass | 11,759 | 6,670 | -43% | 1 | 1 | 0% | 1,594 | 1,547 | -3% | 0 | 0 | — |
case-17 | fail→pass | 11,404 | 2,013 | -82% | 1 | 1 | 0% | 1,724 | 852 | -51% | 0 | 0 | — |
case-08 | pass→pass | 5,512 | 5,166 | -6% | 1 | 1 | 0% | 798 | 1,348 | +69% | 0 | 0 | — |
case-09 | pass→pass | 8,646 | 6,355 | -26% | 1 | 1 | 0% | 1,161 | 1,347 | +16% | 0 | 0 | — |
case-10 | fail→pass | 8,699 | 3,040 | -65% | 1 | 1 | 0% | 1,357 | 973 | -28% | 0 | 0 | — |
case-11 | pass→pass | 15,301 | 3,009 | -80% | 1 | 1 | 0% | 2,003 | 893 | -55% | 0 | 0 | — |
case-12 | pass→pass | 12,206 | 6,913 | -43% | 1 | 1 | 0% | 2,196 | 1,859 | -15% | 0 | 0 | — |
case-13 | fail→pass | 10,449 | 3,238 | -69% | 1 | 1 | 0% | 1,695 | 1,061 | -37% | 0 | 0 | — |
case-14 | fail→pass | 11,342 | 8,167 | -28% | 1 | 1 | 0% | 2,063 | 1,676 | -19% | 0 | 0 | — |
case-15 | pass→pass | 8,855 | 7,696 | -13% | 1 | 1 | 0% | 1,389 | 1,382 | -1% | 0 | 0 | — |
case-16 | fail→pass | 17,273 | 2,701 | -84% | 1 | 1 | 0% | 2,263 | 972 | -57% | 0 | 0 | — |
case-18 | pass→pass | 18,198 | 4,120 | -77% | 1 | 1 | 0% | 1,870 | 1,213 | -35% | 0 | 0 | — |
case-19 | fail→pass | 14,930 | 3,181 | -79% | 1 | 1 | 0% | 1,938 | 1,042 | -46% | 0 | 0 | — |
case-20 | fail→pass | 2,360 | 12,837 | +444% | 1 | 1 | 0% | 282 | 1,873 | +564% | 0 | 0 | — |
case-21 | pass→pass | 32,343 | 34,245 | +6% | 1 | 1 | 0% | 6,577 | 5,901 | -10% | 0 | 0 | — |
case-22 | pass→fail | 6,356 | 12,243 | +93% | 1 | 1 | 0% | 1,057 | 1,378 | +30% | 0 | 0 | — |
DecimalAI ran this skill against gemini-3.6-flash twice over the same eval suite — once with the skill loaded and once without — and compared the two runs case by case. 22 cases were attempted, and 20 counted toward the lift figure. The other 2 produced results that are not comparable between the two arms, so they are excluded from the headline rather than averaged into it. The headline lift of +36 percentage points is the difference between those two pass rates over the 20 comparable cases. 1 case got worse with the skill loaded, and it is included in that figure.
Without the skill loaded, the model failed this case. With it loaded, the same prompt on the same model passed. This is one improved case from the latest verified run; every case, including any that regressed, is in the table above.
Other measured skills in the registry, with their headline benchmark lift.