Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Fetch and summarize review feedback and conversation from a GitHub PR (unresolved review threads, review bodies, and PR conversation comments) without making changes. Use when the user asks to "fetch PR comments", "show PR comments", "check PR for unresolved comments", "list review comments", "what comments are on the PR", "show unresolved threads", or "summarize PR feedback".
.claude/skills/tobihagemann-fetch-pr-comments/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-10 | ✗→✓ | ▲ Improved | 23% | 0% |
| case-16 | ✗→✓ | ▲ Improved | -27% | 0% |
| case-21 | ✗→✓ | ▲ Improved | 303% | 0% |
| case-22 | ✗→✓ | ▲ Improved | -29% | 0% |
| case-18 | ✓→✗ | ▼ Worse | 254% | 0% |
Fetch unresolved review comments, top-level review body comments, and PR conversation comments from a GitHub PR and present them in a readable summary. This is a read-only skill -- it does not evaluate, fix, or reply to any comments.
Auto-detect owner, repo, and PR number from current branch if not provided. Then run scripts/fetch-pr-data.sh, which handles full pagination (review threads, inner comment pages for long threads, reviews, issue comments) and emits a single merged JSON document:
bashbash <skill-dir>/scripts/fetch-pr-data.sh <owner> <repo> <pr_number>
Output shape:
jsonc{ "meta": { "title", "url", "headRefName", "baseRefName" }, "reviewThreads": [ { "id", "isResolved", "isOutdated", "comments": { "nodes": [ { "author", "body", "path", "line", "originalLine", "diffHunk" } ] } } ], "reviews": [ { "author", "body", "state" } ], "issueComments": [ { "author", "body", "createdAt", "url" } ] }
Filter review threads to unresolved only. Filter reviews to those with a non-empty body, excluding PENDING state (unsubmitted drafts). Filter issue comments to those with a non-empty body.
Display a summary header followed by comments grouped by file.
Summary header:
head -> baseTop-level review comments (if any):
Show reviews with non-empty body before the file-grouped threads:
## Review Comments
### @reviewer (CHANGES_REQUESTED)
> Review body text here
### @another-reviewer (COMMENTED)
> Another review body hereIssue comments (if any):
Show PR conversation comments after review comments, ordered by createdAt:
## Issue Comments
### @commenter (2026-04-20)
> Issue comment body here
### @another-commenter (2026-04-21)
> Another issue comment body hereInline threads grouped by file:
For each file with unresolved threads, show:
`## `path/to/file.ts` ### Line 42 (by @reviewer)
<diffHunk from first comment>
> Comment body here
### Line 10 (by @another-reviewer) [outdated]<diffHunk from first comment>
> First comment body
>
> **@reply-author:** Reply bodyFormatting rules:
createdAtdiffHunk from the first comment in each thread as a fenced diff code block before the comment body. This is the code context the reviewer was looking at.line if available. Fall back to originalLine for outdated comments where line is null.[outdated]Then call update_plan to mark this step completed and continue with the next step of the active workflow.
$resolve-pr-comments.| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→fail | 5,867 | 4,457 | -24% | 1 | 1 | 0% | 903 | 1,150 | +27% | 0 | 0 | — |
case-02 | fail→fail | 12,090 | 5,267 | -56% | 1 | 1 | 0% | 2,168 | 1,166 | -46% | 0 | 0 | — |
case-03 | fail→fail | 10,424 | 6,017 | -42% | 1 | 1 | 0% | 1,703 | 1,281 | -25% | 0 | 0 | — |
case-04 | pass→pass | 7,256 | 2,159 | -70% | 1 | 1 | 0% | 1,211 | 1,238 | +2% | 0 | 0 | — |
case-05 | pass→pass | 2,608 | 1,452 | -44% | 1 | 1 | 0% | 411 | 1,106 | +169% | 0 | 0 | — |
case-06 | pass→pass | 7,294 | 2,106 | -71% | 1 | 1 | 0% | 1,187 | 1,240 | +4% | 0 | 0 | — |
case-07 | pass→pass | 11,273 | 2,477 | -78% | 1 | 1 | 0% | 1,737 | 1,296 | -25% | 0 | 0 | — |
case-08 | pass→pass | 3,397 | 2,197 | -35% | 1 | 1 | 0% | 555 | 1,213 | +119% | 0 | 0 | — |
case-09 | fail→fail | 10,011 | 2,900 | -71% | 1 | 1 | 0% | 1,604 | 1,407 | -12% | 0 | 0 | — |
case-10 | fail→pass | 6,304 | 2,408 | -62% | 1 | 1 | 0% | 1,037 | 1,279 | +23% | 0 | 0 | — |
case-11 | pass→pass | 6,248 | 2,067 | -67% | 1 | 1 | 0% | 940 | 1,173 | +25% | 0 | 0 | — |
case-12 | pass→pass | 12,422 | 2,938 | -76% | 1 | 1 | 0% | 1,931 | 1,347 | -30% | 0 | 0 | — |
case-13 | pass→pass | 11,587 | 3,545 | -69% | 1 | 1 | 0% | 1,984 | 1,466 | -26% | 0 | 0 | — |
case-14 | pass→pass | 1,933 | 1,884 | -3% | 1 | 1 | 0% | 333 | 1,214 | +265% | 0 | 0 | — |
case-15 | pass→pass | 9,592 | 1,805 | -81% | 1 | 1 | 0% | 1,573 | 1,127 | -28% | 0 | 0 | — |
case-16 | fail→pass | 10,872 | 2,248 | -79% | 1 | 1 | 0% | 1,651 | 1,201 | -27% | 0 | 0 | — |
case-17 | pass→pass | 3,598 | 6,201 | +72% | 1 | 1 | 0% | 577 | 1,923 | +233% | 0 | 0 | — |
case-18 | pass→fail | 3,334 | 9,525 | +186% | 1 | 1 | 0% | 458 | 1,621 | +254% | 0 | 0 | — |
case-19 | pass→pass | 4,814 | 9,764 | +103% | 1 | 1 | 0% | 834 | 2,521 | +202% | 0 | 0 | — |
case-20 | fail→fail | 10,254 | 1,333 | -87% | 1 | 1 | 0% | 1,661 | 1,057 | -36% | 0 | 0 | — |
case-21 | fail→pass | 2,414 | 1,769 | -27% | 1 | 1 | 0% | 285 | 1,149 | +303% | 0 | 0 | — |
case-22 | fail→pass | 10,472 | 2,168 | -79% | 1 | 1 | 0% | 1,628 | 1,160 | -29% | 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 18 counted toward the lift figure. The other 4 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 +14 percentage points is the difference between those two pass rates over the 18 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.