Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Guide for creating effective skills. Use when creating or updating a skill.
.claude/skills/holaboss-ai-skill-creator/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-05 | ✗→✓ | ▲ Improved | -31% | 0% |
| case-04 | ✗→✓ | ▲ Improved | -64% | 0% |
| case-15 | ✗→✓ | ▲ Improved | -18% | 0% |
| case-21 | ✗→✓ | ▲ Improved | -21% | 0% |
| case-22 | ✓→✗ | ▼ Worse | 38% | 0% |
Use this skill when defining or updating reusable Codex skills.
skills/ at the workspace root.skills/<skill-id>/ and save SKILL.md plus any helper files there.runtime/harnesses/src/embedded-skills/.SKILL.md format:name: <skill-id> (must exactly match the directory name under skills/)description: <one-line summary>markdown--- name: <skill-id> description: <one-line summary> --- # <Readable Skill Title> ## When To Use - ... ## Workflow 1. ... 2. ... ## Examples - ...
SKILL.md concise; use references/scripts only when needed.| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-05 | fail→pass | 14,492 | 10,458 | -28% | 1 | 1 | 0% | 2,799 | 1,945 | -31% | 0 | 0 | — |
case-06 | fail→fail | 12,680 | 12,905 | +2% | 1 | 1 | 0% | 1,997 | 2,770 | +39% | 0 | 0 | — |
case-03 | fail→fail | 8,101 | 6,162 | -24% | 1 | 1 | 0% | 1,371 | 704 | -49% | 0 | 0 | — |
case-04 | fail→pass | 13,853 | 3,357 | -76% | 1 | 1 | 0% | 2,149 | 777 | -64% | 0 | 0 | — |
case-01 | fail→fail | 21,173 | 5,473 | -74% | 1 | 1 | 0% | 3,828 | 686 | -82% | 0 | 0 | — |
case-02 | fail→fail | 18,164 | 6,163 | -66% | 1 | 1 | 0% | 2,790 | 1,330 | -52% | 0 | 0 | — |
case-07 | fail→fail | 20,783 | 11,037 | -47% | 1 | 1 | 0% | 3,039 | 2,345 | -23% | 0 | 0 | — |
case-08 | fail→fail | 3,775 | 3,146 | -17% | 1 | 1 | 0% | 549 | 838 | +53% | 0 | 0 | — |
case-09 | pass→pass | 12,346 | 8,707 | -29% | 1 | 1 | 0% | 1,998 | 1,541 | -23% | 0 | 0 | — |
case-10 | pass→pass | 7,612 | 7,562 | -1% | 1 | 1 | 0% | 1,307 | 1,751 | +34% | 0 | 0 | — |
case-11 | fail→fail | 7,778 | 39,282 | +405% | 1 | 1 | 0% | 1,027 | 6,922 | +574% | 0 | 0 | — |
case-12 | pass→pass | 13,211 | 6,840 | -48% | 1 | 1 | 0% | 2,128 | 1,464 | -31% | 0 | 0 | — |
case-13 | pass→pass | 2,578 | 5,031 | +95% | 1 | 1 | 0% | 450 | 1,029 | +129% | 0 | 0 | — |
case-14 | pass→pass | 17,670 | 10,218 | -42% | 1 | 1 | 0% | 3,024 | 1,926 | -36% | 0 | 0 | — |
case-15 | fail→pass | 6,523 | 3,089 | -53% | 1 | 1 | 0% | 1,007 | 830 | -18% | 0 | 0 | — |
case-16 | pass→pass | 33,421 | 15,003 | -55% | 1 | 1 | 0% | 2,729 | 3,451 | +26% | 0 | 0 | — |
case-17 | pass→pass | 10,464 | 10,857 | +4% | 1 | 1 | 0% | 1,663 | 2,309 | +39% | 0 | 0 | — |
case-18 | pass→pass | 12,287 | 8,105 | -34% | 1 | 1 | 0% | 1,878 | 1,540 | -18% | 0 | 0 | — |
case-19 | pass→pass | 4,777 | 2,533 | -47% | 1 | 1 | 0% | 740 | 630 | -15% | 0 | 0 | — |
case-20 | pass→pass | 11,177 | 5,401 | -52% | 1 | 1 | 0% | 1,493 | 1,123 | -25% | 0 | 0 | — |
case-21 | fail→pass | 8,512 | 3,698 | -57% | 1 | 1 | 0% | 1,138 | 895 | -21% | 0 | 0 | — |
case-22 | pass→fail | 4,482 | 3,702 | -17% | 1 | 1 | 0% | 560 | 771 | +38% | 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. The headline lift of +14 percentage points is the difference between those two pass rates over the 22 comparable cases. 2 cases got worse with the skill loaded, and they are 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.