Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Check for pull requests requesting your review and selectively review them using the review-pr skill
.claude/skills/nudgebee-pr-backlog/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-08 | ✗→✓ | ▲ Improved | -12% | 0% |
| case-10 | ✗→✓ | ▲ Improved | -23% | 0% |
| case-11 | ✗→✓ | ▲ Improved | -25% | 0% |
| case-16 | ✗→✓ | ▲ Improved | -29% | 0% |
| case-17 | ✗→✓ | ▲ Improved | 9% | 0% |
Check for pull requests where you are requested as a reviewer, display them in a list, and let you select which ones to review.
Use the GitHub CLI to fetch all open, non-draft PRs where you are requested as a reviewer:
bashgh pr list --search "review-requested:@me state:open draft:false" --json number,title,author,createdAt,updatedAt,url --limit 50
This returns JSON with PR details. Store this information.
Parse the JSON output and display the PRs in a clear, readable format:
Found {N} PRs requesting your review:
1. PR #{number}: {title}
Author: {author.login}
Created: {createdAt}
URL: {url}
2. PR #{number}: {title}
Author: {author.login}
Created: {createdAt}
URL: {url}
...Important: If no PRs are found, display: "No PRs currently requesting your review. ✅"
Present options to the user using the AskUserQuestion tool:
json{ "questions": [{ "question": "Which PRs would you like to review?", "header": "Select PRs", "multiSelect": true, "options": [ { "label": "PR #{number}: {short-title}", "description": "By {author} - {relative-time}" }, ... ] }] }
Note: Limit the displayed title to 50 characters in the label. If there are more than 4 PRs, show the first 4 and add an option "Show all" or let the user provide a specific PR number.
For each selected PR number, invoke the review-pr skill:
bash# For each selected PR /review-pr {pr_number}
Implementation: Use the Skill tool to invoke review-pr:
Skill tool with:
skill: "review-pr"
args: "{pr_number}"After all reviews are complete, display a summary:
Review session complete!
Reviewed PRs:
- PR #{number}: {title} ✅
- PR #{number}: {title} ✅
- ...
Total: {N} PRs reviewedIf the user provides arguments like /pr-backlog all, skip the selection step and review ALL PRs requesting your review automatically (use with caution).
gh CLI is not installed or not authenticated, display: "GitHub CLI (gh) is required. Please install and authenticate: gh auth login"| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-08 | fail→pass | 6,541 | 2,225 | -66% | 1 | 1 | 0% | 1,243 | 1,091 | -12% | 0 | 0 | — |
case-13 | fail→fail | 9,141 | 2,816 | -69% | 1 | 1 | 0% | 1,587 | 1,198 | -25% | 0 | 0 | — |
case-19 | pass→pass | 13,507 | 1,938 | -86% | 1 | 1 | 0% | 2,187 | 1,016 | -54% | 0 | 0 | — |
case-01 | fail→fail | 7,428 | 9,583 | +29% | 1 | 1 | 0% | 1,089 | 923 | -15% | 0 | 0 | — |
case-02 | fail→fail | 5,749 | 5,000 | -13% | 1 | 1 | 0% | 799 | 1,007 | +26% | 0 | 0 | — |
case-03 | fail→fail | 6,328 | 4,369 | -31% | 1 | 1 | 0% | 990 | 878 | -11% | 0 | 0 | — |
case-04 | pass→fail | 4,705 | 7,770 | +65% | 1 | 1 | 0% | 739 | 1,217 | +65% | 0 | 0 | — |
case-05 | fail→fail | 3,274 | 6,961 | +113% | 1 | 1 | 0% | 402 | 1,074 | +167% | 0 | 0 | — |
case-06 | pass→fail | 5,149 | 6,493 | +26% | 1 | 1 | 0% | 808 | 973 | +20% | 0 | 0 | — |
case-07 | pass→fail | 5,550 | 6,661 | +20% | 1 | 1 | 0% | 868 | 1,162 | +34% | 0 | 0 | — |
case-09 | pass→pass | 9,192 | 4,717 | -49% | 1 | 1 | 0% | 1,607 | 1,462 | -9% | 0 | 0 | — |
case-10 | fail→pass | 12,451 | 4,849 | -61% | 1 | 1 | 0% | 2,020 | 1,565 | -23% | 0 | 0 | — |
case-11 | fail→pass | 10,122 | 2,897 | -71% | 1 | 1 | 0% | 1,525 | 1,148 | -25% | 0 | 0 | — |
case-12 | pass→pass | 11,036 | 2,186 | -80% | 1 | 1 | 0% | 1,651 | 1,001 | -39% | 0 | 0 | — |
case-14 | pass→pass | 6,578 | 2,493 | -62% | 1 | 1 | 0% | 1,042 | 1,078 | +3% | 0 | 0 | — |
case-15 | pass→pass | 9,501 | 4,082 | -57% | 1 | 1 | 0% | 1,747 | 1,369 | -22% | 0 | 0 | — |
case-16 | fail→pass | 9,549 | 1,918 | -80% | 1 | 1 | 0% | 1,396 | 985 | -29% | 0 | 0 | — |
case-17 | fail→pass | 7,608 | 3,305 | -57% | 1 | 1 | 0% | 1,152 | 1,254 | +9% | 0 | 0 | — |
case-18 | fail→pass | 12,390 | 3,457 | -72% | 1 | 1 | 0% | 1,962 | 1,323 | -33% | 0 | 0 | — |
case-20 | fail→pass | 3,827 | 2,206 | -42% | 1 | 1 | 0% | 612 | 1,060 | +73% | 0 | 0 | — |
case-21 | pass→pass | 5,942 | 1,507 | -75% | 1 | 1 | 0% | 1,047 | 934 | -11% | 0 | 0 | — |
case-22 | pass→pass | 9,629 | 2,671 | -72% | 1 | 1 | 0% | 1,510 | 1,097 | -27% | 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 15 counted toward the lift figure. The other 7 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 +18 percentage points is the difference between those two pass rates over the 15 comparable cases. 4 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.