Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Cross-phase audit of all outstanding UAT and verification items
.claude/skills/coco-research-gsd-audit-uat/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-07 | ✗→✓ | ▲ Improved | -84% | 0% |
| case-09 | ✗→✓ | ▲ Improved | -60% | 0% |
| case-15 | ✗→✓ | ▲ Improved | -19% | 0% |
| case-12 | ✗→✓ | ▲ Improved | -35% | 0% |
| case-17 | ✗→✓ | ▲ Improved | -58% | 0% |
<objective> Scan all phases for pending, skipped, blocked, and human_needed UAT items. Cross-reference against codebase to detect stale documentation. Produce prioritized human test plan. </objective>
<execution_context> @$HOME/.claude/get-shit-done/workflows/audit-uat.md </execution_context>
<context> Core planning files are loaded in-workflow via CLI.
Scope: Glob: .planning/phases//-UAT.md Glob: .planning/phases//-VERIFICATION.md </context>
<process> Execute the audit-uat workflow from @$HOME/.claude/get-shit-done/workflows/audit-uat.md end-to-end. </process>
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-04 | fail→fail | 15,406 | 11,578 | -25% | 1 | 1 | 0% | 144 | 525 | +265% | 0 | 0 | — |
case-05 | fail→fail | 7,738 | 20,238 | +162% | 1 | 1 | 0% | 270 | 877 | +225% | 0 | 0 | — |
case-01 | fail→fail | 11,756 | 10,967 | -7% | 1 | 1 | 0% | 266 | 357 | +34% | 0 | 0 | — |
case-02 | fail→fail | 2,569 | 15,684 | +511% | 1 | 1 | 0% | 278 | 436 | +57% | 0 | 0 | — |
case-03 | fail→fail | 15,191 | 11,906 | -22% | 1 | 1 | 0% | 1,591 | 343 | -78% | 0 | 0 | — |
case-06 | pass→fail | 15,567 | 13,453 | -14% | 1 | 1 | 0% | 2,205 | 815 | -63% | 0 | 0 | — |
case-07 | fail→pass | 21,167 | 7,500 | -65% | 1 | 1 | 0% | 2,407 | 394 | -84% | 0 | 0 | — |
case-08 | pass→pass | 11,320 | 9,962 | -12% | 1 | 1 | 0% | 1,805 | 1,003 | -44% | 0 | 0 | — |
case-09 | fail→pass | 12,308 | 2,416 | -80% | 1 | 1 | 0% | 1,079 | 430 | -60% | 0 | 0 | — |
case-15 | fail→pass | 9,342 | 11,132 | +19% | 1 | 1 | 0% | 1,657 | 1,346 | -19% | 0 | 0 | — |
case-10 | fail→fail | 14,257 | 16,372 | +15% | 1 | 1 | 0% | 1,643 | 622 | -62% | 0 | 0 | — |
case-11 | fail→fail | 16,370 | 12,811 | -22% | 1 | 1 | 0% | 1,939 | 347 | -82% | 0 | 0 | — |
case-12 | fail→pass | 11,922 | 5,547 | -53% | 1 | 1 | 0% | 1,268 | 827 | -35% | 0 | 0 | — |
case-13 | pass→pass | 13,471 | 21,223 | +58% | 1 | 1 | 0% | 2,317 | 3,000 | +29% | 0 | 0 | — |
case-14 | pass→pass | 15,019 | 7,165 | -52% | 1 | 1 | 0% | 1,382 | 1,077 | -22% | 0 | 0 | — |
case-16 | pass→pass | 11,660 | 7,608 | -35% | 1 | 1 | 0% | 1,676 | 1,276 | -24% | 0 | 0 | — |
case-17 | fail→pass | 15,120 | 7,815 | -48% | 1 | 1 | 0% | 1,594 | 669 | -58% | 0 | 0 | — |
case-18 | fail→pass | 21,483 | 16,188 | -25% | 1 | 1 | 0% | 2,269 | 2,006 | -12% | 0 | 0 | — |
case-19 | pass→fail | 6,800 | 16,890 | +148% | 1 | 1 | 0% | 1,059 | 554 | -48% | 0 | 0 | — |
case-20 | fail→pass | 8,605 | 9,563 | +11% | 1 | 1 | 0% | 1,218 | 1,382 | +13% | 0 | 0 | — |
case-21 | fail→pass | 11,544 | 7,415 | -36% | 1 | 1 | 0% | 850 | 1,218 | +43% | 0 | 0 | — |
case-22 | pass→pass | 17,728 | 14,119 | -20% | 1 | 1 | 0% | 1,684 | 1,302 | -23% | 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 13 counted toward the lift figure. The other 9 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 +27 percentage points is the difference between those two pass rates over the 13 comparable cases. 3 cases got worse with the skill loaded, and they are included in that figure.
The publisher has shipped newer versions since this run, so these numbers describe v1, not the version currently listed.
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.