Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Inspect unresolved GitHub pull request review threads, group actionable feedback, implement selected fixes locally, and only reply or resolve threads after explicit user intent.
.claude/skills/devin-axis-github-review-follow-up/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-13 | ✗→✓ | ▲ Improved | 266% | 0% |
| case-17 | ✗→✓ | ▲ Improved | -9% | 0% |
| case-12 | ✓→✗ | ▼ Worse | -13% | 0% |
| case-18 | ✓→✗ | ▼ Worse | -63% | 0% |
| case-19 | ✓→✓ | ＝ Same ✓ | 98% | 0% |
Use this skill for requested changes, inline review comments, and unresolved Pull Request feedback.
owner, repo, and PR number from the request, URL, or current branch.pull-request-detail and review-threads actions.resolve-review-thread unless the user explicitly asks.| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-16 | fail→fail | 14,632 | 13,318 | -9% | 1 | 1 | 0% | 169 | 474 | +180% | 0 | 0 | — |
case-01 | fail→fail | 20,337 | 16,474 | -19% | 1 | 1 | 0% | 1,148 | 608 | -47% | 0 | 0 | — |
case-02 | fail→fail | 15,281 | 39,075 | +156% | 1 | 1 | 0% | 1,543 | 786 | -49% | 0 | 0 | — |
case-03 | fail→fail | 16,763 | 15,855 | -5% | 1 | 1 | 0% | 327 | 661 | +102% | 0 | 0 | — |
case-04 | fail→fail | 15,793 | 17,755 | +12% | 1 | 1 | 0% | 1,256 | 1,585 | +26% | 0 | 0 | — |
case-10 | fail→fail | 15,761 | 19,725 | +25% | 1 | 1 | 0% | 268 | 724 | +170% | 0 | 0 | — |
case-05 | fail→fail | 18,384 | 15,036 | -18% | 1 | 1 | 0% | 2,473 | 563 | -77% | 0 | 0 | — |
case-06 | fail→fail | 9,965 | 11,923 | +20% | 1 | 1 | 0% | 863 | 1,422 | +65% | 0 | 0 | — |
case-07 | fail→fail | 7,580 | 17,400 | +130% | 1 | 1 | 0% | 391 | 620 | +59% | 0 | 0 | — |
case-08 | fail→fail | 16,201 | 41,228 | +154% | 1 | 1 | 0% | 306 | 6,033 | +1872% | 0 | 0 | — |
case-09 | fail→fail | 26,523 | 15,885 | -40% | 1 | 1 | 0% | 1,650 | 544 | -67% | 0 | 0 | — |
case-11 | fail→fail | 13,646 | 20,784 | +52% | 1 | 1 | 0% | 1,347 | 689 | -49% | 0 | 0 | — |
case-12 | pass→fail | 26,617 | 22,113 | -17% | 1 | 1 | 0% | 1,821 | 1,584 | -13% | 0 | 0 | — |
case-13 | fail→pass | 10,065 | 29,042 | +189% | 1 | 1 | 0% | 705 | 2,579 | +266% | 0 | 0 | — |
case-14 | fail→fail | 14,610 | 21,311 | +46% | 1 | 1 | 0% | 168 | 817 | +386% | 0 | 0 | — |
case-15 | fail→fail | 16,241 | 16,839 | +4% | 1 | 1 | 0% | 1,807 | 2,410 | +33% | 0 | 0 | — |
case-17 | fail→pass | 10,983 | 8,677 | -21% | 1 | 1 | 0% | 1,000 | 906 | -9% | 0 | 0 | — |
case-18 | pass→fail | 13,272 | 18,267 | +38% | 1 | 1 | 0% | 1,342 | 495 | -63% | 0 | 0 | — |
case-19 | pass→pass | 9,336 | 11,277 | +21% | 1 | 1 | 0% | 655 | 1,300 | +98% | 0 | 0 | — |
case-20 | fail→fail | 18,699 | 17,799 | -5% | 1 | 1 | 0% | 431 | 582 | +35% | 0 | 0 | — |
case-21 | fail→fail | 19,806 | 15,309 | -23% | 1 | 1 | 0% | 1,095 | 531 | -52% | 0 | 0 | — |
case-22 | fail→fail | 15,455 | 17,787 | +15% | 1 | 1 | 0% | 233 | 670 | +188% | 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 6 counted toward the lift figure. The other 16 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 0 percentage points is the difference between those two pass rates over the 6 comparable cases. 6 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.