Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Monitor PR checks and fix failures until green. Uses gh pr checks as the source of truth for PR-attached checks.
.claude/skills/kunanonj-cursor-plugin-teamkit-loop-on-ci/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-07 | ✗→✓ | ▲ Improved | -42% | 0% |
| case-15 | ✗→✓ | ▲ Improved | -5% | 0% |
| case-17 | ✗→✓ | ▲ Improved | 2% | 0% |
| case-18 | ✓→✗ | ▼ Worse | -31% | 0% |
| case-04 | ✓→✓ | ＝ Same ✓ | -41% | 0% |
Need to watch a branch or pull request and iterate on CI failures until all required checks are green.
Use gh pr checks as the source of truth. It includes all PR-attached checks, while gh run list only covers GitHub Actions.
gh pr checks --watch --fail-fast.bash# Resolve the active PR gh pr view --json number,url,headRefName # Inspect all attached checks gh pr checks --json name,bucket,state,workflow,link # Watch pending checks and fail fast gh pr checks --watch --fail-fast # GitHub Actions logs, when the failing check links to a GHA run gh run view <run-id> --log-failed
--no-verify) to force progress.gh pr checks --json name,bucket,state,workflow,link after every push; the check set can change.| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→fail | 11,016 | 3,913 | -64% | 1 | 1 | 0% | 1,695 | 526 | -69% | 0 | 0 | — |
case-02 | fail→fail | 6,037 | 3,564 | -41% | 1 | 1 | 0% | 383 | 612 | +60% | 0 | 0 | — |
case-03 | fail→fail | 5,482 | 3,212 | -41% | 1 | 1 | 0% | 588 | 560 | -5% | 0 | 0 | — |
case-04 | pass→pass | 8,839 | 3,277 | -63% | 1 | 1 | 0% | 1,691 | 996 | -41% | 0 | 0 | — |
case-05 | pass→pass | 4,377 | 2,368 | -46% | 1 | 1 | 0% | 864 | 749 | -13% | 0 | 0 | — |
case-06 | pass→pass | 3,408 | 1,516 | -56% | 1 | 1 | 0% | 612 | 631 | +3% | 0 | 0 | — |
case-07 | fail→pass | 10,170 | 3,510 | -65% | 1 | 1 | 0% | 1,655 | 964 | -42% | 0 | 0 | — |
case-08 | pass→pass | 7,019 | 3,085 | -56% | 1 | 1 | 0% | 1,267 | 777 | -39% | 0 | 0 | — |
case-09 | pass→pass | 10,520 | 5,927 | -44% | 1 | 1 | 0% | 1,787 | 1,424 | -20% | 0 | 0 | — |
case-10 | pass→pass | 10,469 | 4,138 | -60% | 1 | 1 | 0% | 1,708 | 1,103 | -35% | 0 | 0 | — |
case-11 | pass→pass | 8,334 | 3,875 | -54% | 1 | 1 | 0% | 1,357 | 929 | -32% | 0 | 0 | — |
case-12 | pass→pass | 5,564 | 2,133 | -62% | 1 | 1 | 0% | 1,102 | 793 | -28% | 0 | 0 | — |
case-13 | pass→pass | 7,988 | 3,410 | -57% | 1 | 1 | 0% | 1,580 | 1,037 | -34% | 0 | 0 | — |
case-14 | pass→pass | 9,153 | 1,873 | -80% | 1 | 1 | 0% | 1,589 | 695 | -56% | 0 | 0 | — |
case-15 | fail→pass | 13,421 | 9,611 | -28% | 1 | 1 | 0% | 2,219 | 2,099 | -5% | 0 | 0 | — |
case-16 | pass→pass | 2,767 | 1,225 | -56% | 1 | 1 | 0% | 527 | 597 | +13% | 0 | 0 | — |
case-17 | fail→pass | 4,978 | 2,045 | -59% | 1 | 1 | 0% | 741 | 758 | +2% | 0 | 0 | — |
case-18 | pass→fail | 8,508 | 3,177 | -63% | 1 | 1 | 0% | 1,322 | 911 | -31% | 0 | 0 | — |
case-19 | pass→pass | 10,533 | 10,686 | +1% | 1 | 1 | 0% | 1,814 | 1,114 | -39% | 0 | 0 | — |
case-20 | pass→pass | 6,738 | 4,534 | -33% | 1 | 1 | 0% | 1,291 | 1,195 | -7% | 0 | 0 | — |
case-21 | pass→pass | 5,586 | 4,798 | -14% | 1 | 1 | 0% | 1,207 | 1,244 | +3% | 0 | 0 | — |
case-22 | pass→pass | 5,037 | 4,064 | -19% | 1 | 1 | 0% | 1,008 | 1,034 | +3% | 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 19 counted toward the lift figure. The other 3 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 +9 percentage points is the difference between those two pass rates over the 19 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.