Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Address review and issue comments on the open GitHub PR for the current branch using gh CLI. Use when user says "address PR comments", "fix review feedback", "respond to PR review", or "handle PR comments". Verifies gh auth first and prompts to authenticate if not logged in. Do NOT use for creating PRs, CI debugging (use gh-fix-ci), or general Git operations.
.claude/skills/tech-leads-club-gh-address-comments/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-06 | ✗→✓ | ▲ Improved | -79% | 0% |
| case-08 | ✗→✓ | ▲ Improved | -42% | 0% |
| case-11 | ✗→✓ | ▲ Improved | -59% | 0% |
| case-07 | ✓→✗ | ▼ Worse | -68% | 0% |
| case-10 | ✓→✗ | ▼ Worse | -55% | 0% |
Guide to find the open PR for the current branch and address its comments with gh CLI.
Prerequisites: Ensure gh is authenticated before running commands. Check authentication status with gh auth status. If not authenticated, instruct the user to run gh auth login to authenticate with GitHub.
Notes:
gh auth login, then retry.| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-04 | fail→fail | 6,644 | 2,749 | -59% | 1 | 1 | 0% | 1,000 | 636 | -36% | 0 | 0 | — |
case-01 | fail→fail | 4,590 | 4,937 | +8% | 1 | 1 | 0% | 651 | 463 | -29% | 0 | 0 | — |
case-02 | fail→fail | 4,299 | 4,393 | +2% | 1 | 1 | 0% | 715 | 402 | -44% | 0 | 0 | — |
case-03 | fail→fail | 5,032 | 5,467 | +9% | 1 | 1 | 0% | 684 | 348 | -49% | 0 | 0 | — |
case-05 | pass→pass | 5,860 | 2,418 | -59% | 1 | 1 | 0% | 995 | 626 | -37% | 0 | 0 | — |
case-06 | fail→pass | 11,810 | 1,359 | -88% | 1 | 1 | 0% | 1,854 | 382 | -79% | 0 | 0 | — |
case-07 | pass→fail | 11,641 | 2,553 | -78% | 1 | 1 | 0% | 1,881 | 607 | -68% | 0 | 0 | — |
case-08 | fail→pass | 7,445 | 3,414 | -54% | 1 | 1 | 0% | 1,174 | 686 | -42% | 0 | 0 | — |
case-09 | fail→fail | 4,632 | 63,197 | +1264% | 1 | 1 | 0% | 747 | 496 | -34% | 0 | 0 | — |
case-10 | pass→fail | 8,632 | 3,007 | -65% | 1 | 1 | 0% | 1,474 | 662 | -55% | 0 | 0 | — |
case-11 | fail→pass | 10,537 | 2,836 | -73% | 1 | 1 | 0% | 1,770 | 730 | -59% | 0 | 0 | — |
case-12 | fail→fail | 8,545 | 2,891 | -66% | 1 | 1 | 0% | 1,403 | 570 | -59% | 0 | 0 | — |
case-13 | pass→pass | 6,169 | 2,986 | -52% | 1 | 1 | 0% | 1,027 | 699 | -32% | 0 | 0 | — |
case-14 | fail→fail | 10,965 | 2,372 | -78% | 1 | 1 | 0% | 1,710 | 627 | -63% | 0 | 0 | — |
case-15 | fail→fail | 6,342 | 6,914 | +9% | 1 | 1 | 0% | 983 | 574 | -42% | 0 | 0 | — |
case-16 | pass→pass | 1,521 | 1,326 | -13% | 1 | 1 | 0% | 195 | 351 | +80% | 0 | 0 | — |
case-17 | pass→fail | 10,562 | 2,055 | -81% | 1 | 1 | 0% | 1,528 | 476 | -69% | 0 | 0 | — |
case-18 | fail→fail | 2,420 | 6,941 | +187% | 1 | 1 | 0% | 359 | 1,413 | +294% | 0 | 0 | — |
case-19 | pass→pass | 5,348 | 2,707 | -49% | 1 | 1 | 0% | 913 | 615 | -33% | 0 | 0 | — |
case-20 | pass→pass | 4,948 | 3,809 | -23% | 1 | 1 | 0% | 913 | 859 | -6% | 0 | 0 | — |
case-21 | pass→pass | 6,105 | 3,326 | -46% | 1 | 1 | 0% | 1,063 | 822 | -23% | 0 | 0 | — |
case-22 | pass→pass | 10,516 | 4,953 | -53% | 1 | 1 | 0% | 1,784 | 1,021 | -43% | 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 0 percentage points is the difference between those two pass rates over the 19 comparable cases. 3 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.