Install any skill in seconds. Free to start, no credit card required.
Get Started Free →GitHub PR utilities for code review workflows
.claude/skills/aiskillstore-github-pr/SKILL.md| Model | Eval pass | Runs |
|---|---|---|
| gemini-3.6-flash | 100% | 6 |
| gemini-3.1-pro-preview | 100% | 2 |
| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-19 | ✗→✓ | ▲ Improved | 44% | 0% |
| case-08 | ✗→✓ | ▲ Improved | -12% | 0% |
| case-06 | ✗→✓ | ▲ Improved | 123% | 0% |
| case-07 | ✗→✓ | ▲ Improved | -20% | 0% |
| case-13 | ✗→✓ | ▲ Improved | 57% | 0% |
CLI tools for GitHub pull request operations. Designed to support automated code review workflows. Requires the GitHub CLI (gh) to be installed and authenticated.
bash brew install gh gh auth login
Determines if a PR should be reviewed by checking various conditions.
bashbun .opencode/skill/github-pr/check-review-needed.js [pr-number]
Arguments:
pr-number - PR number (optional, defaults to current branch's PR)Output: JSON object with:
shouldReview - boolean indicating if review should proceedreason - explanation for the decisionprNumber - the PR number checkedConditions checked:
Examples:
bash# Check current branch's PR bun .opencode/skill/github-pr/check-review-needed.js # Check specific PR bun .opencode/skill/github-pr/check-review-needed.js 123
Finds AGENTS.md (or CLAUDE.md) files relevant to a PR's changes.
bashbun .opencode/skill/github-pr/list-guideline-files.js [pr-number] [--json]
Arguments:
pr-number - PR number (optional, defaults to current branch's PR)Options:
--json - Output as JSON array with file contentsSearch locations:
Priority: If both AGENTS.md and CLAUDE.md exist in the same directory, AGENTS.md takes precedence.
Examples:
bash# List guideline files for current PR bun .opencode/skill/github-pr/list-guideline-files.js # Get full content as JSON bun .opencode/skill/github-pr/list-guideline-files.js 123 --json
JSON Output Format:
json[ { "path": "AGENTS.md", "content": "# Project Guidelines\n..." }, { "path": "src/components/AGENTS.md", "content": "# Component Guidelines\n..." } ]
Posts a review comment on a specific line or line range in a PR.
bashbun .opencode/skill/github-pr/post-inline-comment.js <pr-number> --path <file> --line <n> --body <text>
Arguments:
pr-number - PR number (optional if on a PR branch)Options:
--path <file> - File path to comment on (required)--line <n> - Line number to comment on (required)--start-line <n> - Start line for multi-line comments (optional)--body <text> - Comment body in markdown (required)Suggestion blocks: Include a suggestion block for small fixes that can be committed directly:
`markdownFix the error handling:
try { await authenticate(); } catch (e) { handleAuthError(e); }
Important: Suggestions must be complete. The author should be able to click "Commit suggestion" without needing additional changes elsewhere.
Examples:
bash# Single line comment bun .opencode/skill/github-pr/post-inline-comment.js 123 \ --path src/auth.ts \ --line 67 \ --body "Missing error handling for OAuth callback" # Multi-line comment (lines 65-70) bun .opencode/skill/github-pr/post-inline-comment.js 123 \ --path src/auth.ts \ --line 70 \ --start-line 65 \ --body "This authentication block needs refactoring"
These tools wrap the GitHub CLI (gh). For operations not covered by these utilities, use gh directly:
bash# View PR details gh pr view 123 --json title,body,state,isDraft,files # Get PR diff gh pr diff 123 # View PR comments gh pr view 123 --comments # Post a regular comment gh pr comment 123 --body "Comment text" # View file at PR head gh api repos/{owner}/{repo}/contents/{path}?ref={branch}
--json flag when you need to process output programmatically| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-19 | fail→pass | 11,591 | 8,682 | -25% | 1 | 1 | 0% | 987 | 1,426 | +44% | 0 | 0 | — |
case-20 | pass→fail | 7,129 | 10,498 | +47% | 1 | 1 | 0% | 1,148 | 1,495 | +30% | 0 | 0 | — |
case-08 | fail→pass | 9,906 | 6,324 | -36% | 1 | 1 | 0% | 1,520 | 1,336 | -12% | 0 | 0 | — |
case-01 | fail→fail | 4,737 | 15,196 | +221% | 1 | 1 | 0% | 691 | 1,407 | +104% | 0 | 0 | — |
case-02 | fail→fail | 5,824 | 5,423 | -7% | 1 | 1 | 0% | 924 | 1,529 | +65% | 0 | 0 | — |
case-03 | fail→fail | 5,155 | 7,174 | +39% | 1 | 1 | 0% | 845 | 1,608 | +90% | 0 | 0 | — |
case-04 | fail→fail | 10,025 | 2,626 | -74% | 1 | 1 | 0% | 758 | 1,456 | +92% | 0 | 0 | — |
case-05 | pass→pass | 13,285 | 7,763 | -42% | 1 | 1 | 0% | 1,232 | 1,571 | +28% | 0 | 0 | — |
case-06 | fail→pass | 19,966 | 7,258 | -64% | 1 | 1 | 0% | 704 | 1,570 | +123% | 0 | 0 | — |
case-07 | fail→pass | 16,377 | 4,653 | -72% | 1 | 1 | 0% | 1,821 | 1,463 | -20% | 0 | 0 | — |
case-13 | fail→pass | 6,190 | 3,033 | -51% | 1 | 1 | 0% | 1,077 | 1,689 | +57% | 0 | 0 | — |
case-09 | pass→pass | 6,741 | 7,563 | +12% | 1 | 1 | 0% | 1,065 | 1,564 | +47% | 0 | 0 | — |
case-10 | fail→pass | 3,144 | 10,812 | +244% | 1 | 1 | 0% | 431 | 1,701 | +295% | 0 | 0 | — |
case-11 | pass→pass | 8,705 | 7,804 | -10% | 1 | 1 | 0% | 1,464 | 1,683 | +15% | 0 | 0 | — |
case-12 | pass→pass | 12,441 | 2,125 | -83% | 1 | 1 | 0% | 1,777 | 1,504 | -15% | 0 | 0 | — |
case-14 | pass→pass | 8,697 | 6,240 | -28% | 1 | 1 | 0% | 514 | 1,340 | +161% | 0 | 0 | — |
case-15 | pass→pass | 4,938 | 7,206 | +46% | 1 | 1 | 0% | 833 | 1,370 | +64% | 0 | 0 | — |
case-16 | pass→fail | 8,521 | 10,913 | +28% | 1 | 1 | 0% | 491 | 1,499 | +205% | 0 | 0 | — |
case-17 | pass→pass | 13,520 | 9,384 | -31% | 1 | 1 | 0% | 1,591 | 2,024 | +27% | 0 | 0 | — |
case-18 | fail→pass | 11,418 | 1,637 | -86% | 1 | 1 | 0% | 1,058 | 1,427 | +35% | 0 | 0 | — |
case-21 | pass→fail | 10,006 | 11,410 | +14% | 1 | 1 | 0% | 843 | 1,481 | +76% | 0 | 0 | — |
case-22 | fail→fail | 10,053 | 20,690 | +106% | 1 | 1 | 0% | 921 | 1,797 | +95% | 0 | 0 | — |
case-23 | fail→pass | 17,873 | 6,891 | -61% | 1 | 1 | 0% | 1,935 | 1,495 | -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. 23 cases were attempted, and 15 counted toward the lift figure. The other 8 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 +22 percentage points is the difference between those two pass rates over the 15 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.