Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Show the current user's open PRs across every pod registered under .github/teams/, grouped by merge readiness, with copy-paste Slack ping messages routed to the owning pod's team. Use when the user asks about their own PRs, merge readiness, who to ping, or invokes /qv-pr-mine.
.claude/skills/tetherto-qv-pr-mine/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-04 | ✗→✓ | ▲ Improved | 7% | 0% |
| case-06 | ✗→✓ | ▲ Improved | -8% | 0% |
| case-07 | ✗→✓ | ▲ Improved | -17% | 0% |
| case-08 | ✗→✓ | ▲ Improved | 43% | 0% |
| case-09 | ✗→✓ | ▲ Improved | -18% | 0% |
Cross-pod skill. Surfaces the current user's open PRs in the configured repo and emits copy-paste Slack ping messages. The owning pod for each PR is auto-detected from the touched files against every .github/teams/<pod>.json's ownedPaths, so per-PR ping logic uses the right team automatically — no --pod argument needed.
Use when:
/qv-pr-minegh CLI installed and authenticated (gh auth status).github/teams/<pod>.json~/.config/qvac-pr-skills/slack.json (auto-bootstrapped on first run; see workflow step 2)bashnode .cursor/skills/_lib/pr-skills/pr-status.mjs --mode my
--pod is intentionally omitted: --mode my discovers every pod and routes each PR to its owning pod automatically.
--mode my. The script uses .cursor/skills/_lib/pr-skills/pr-activity.mjs for shared PR state collection and ~/.config/qvac-pr-skills/config.json for repo/stale settings when available.SLACK_VALIDATION_REQUIRED <N>, run the validation flow before presenting any output to the user:a. Read ~/.config/qvac-pr-skills/slack.json. b. For each login in pendingReview, present the proposed handle to the user via AskQuestion so they can confirm or correct. Use one question per pending login. Show the GitHub login and the proposed handle as the option label, e.g. <github-login> -> <proposed-handle>. Provide options: "keep proposed", "edit (then prompt for new handle)". Do NOT inline any names from the proposed map into commentary, examples, or follow-up text. c. For any logins the user chose to edit, ask one follow-up question per login for the corrected handle as free text. d. Apply corrections to state.map, set state.pendingReview = [], save the file (atomic write). e. Re-run the script. The marker should no longer fire.
/qv-pr-review.".github/teams/<pod>.json ownedPaths; ping logic is skipped (rare; usually means the user's PR is outside the pod system)Each non-empty group of PRs needing pings includes ready-to-copy Slack messages with @-mentions and PR links sourced from ~/.config/qvac-pr-skills/slack.json (falling back to @<github-login> when a handle is not yet mapped).
For each of the user's PRs, the script:
.github/teams/*.json) and picks the first one whose ownedPaths contains any touched path as a prefix. The walk order is readdir order of .github/teams/.leads, members, and Slack handles for the PR's "needs approval" / "ping" semantics.If a PR touches files spanning multiple pods, the first match wins. This is fine for the typical case where a PR is owned by a single team. If two pods commonly share files, rename one of them in .github/teams/ so the desired pod sorts first.
The script auto-fills new entries from gh api users/<login> when:
.github/teams/<pod>.json that is not yet in the map.Newly seeded logins land in pendingReview and trigger the validation gate above on the next --mode my run. Edit ~/.config/qvac-pr-skills/slack.json directly at any time to update handles; the file is per-user and never committed.
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-04 | fail→pass | 9,307 | 2,344 | -75% | 1 | 1 | 0% | 1,462 | 1,561 | +7% | 0 | 0 | — |
case-05 | fail→fail | 8,781 | 4,416 | -50% | 1 | 1 | 0% | 1,312 | 1,984 | +51% | 0 | 0 | — |
case-01 | fail→fail | 5,428 | 3,437 | -37% | 1 | 1 | 0% | 642 | 1,343 | +109% | 0 | 0 | — |
case-02 | fail→fail | 10,005 | 4,103 | -59% | 1 | 1 | 0% | 1,528 | 1,363 | -11% | 0 | 0 | — |
case-03 | fail→fail | 6,596 | 3,732 | -43% | 1 | 1 | 0% | 1,051 | 1,354 | +29% | 0 | 0 | — |
case-06 | fail→pass | 12,584 | 2,935 | -77% | 1 | 1 | 0% | 1,827 | 1,673 | -8% | 0 | 0 | — |
case-07 | fail→pass | 14,036 | 4,133 | -71% | 1 | 1 | 0% | 2,231 | 1,856 | -17% | 0 | 0 | — |
case-08 | fail→pass | 7,139 | 2,258 | -68% | 1 | 1 | 0% | 1,108 | 1,586 | +43% | 0 | 0 | — |
case-09 | fail→pass | 13,092 | 3,271 | -75% | 1 | 1 | 0% | 2,134 | 1,740 | -18% | 0 | 0 | — |
case-10 | fail→pass | 14,192 | 3,288 | -77% | 1 | 1 | 0% | 2,247 | 1,699 | -24% | 0 | 0 | — |
case-11 | fail→pass | 11,267 | 3,198 | -72% | 1 | 1 | 0% | 1,778 | 1,652 | -7% | 0 | 0 | — |
case-12 | fail→fail | 9,568 | 3,646 | -62% | 1 | 1 | 0% | 1,603 | 1,708 | +7% | 0 | 0 | — |
case-13 | fail→pass | 12,155 | 3,678 | -70% | 1 | 1 | 0% | 1,854 | 1,836 | -1% | 0 | 0 | — |
case-14 | fail→pass | 8,805 | 2,337 | -73% | 1 | 1 | 0% | 1,461 | 1,545 | +6% | 0 | 0 | — |
case-15 | fail→fail | 6,070 | 1,737 | -71% | 1 | 1 | 0% | 934 | 1,427 | +53% | 0 | 0 | — |
case-16 | fail→pass | 6,401 | 2,220 | -65% | 1 | 1 | 0% | 965 | 1,466 | +52% | 0 | 0 | — |
case-17 | fail→fail | 13,600 | 2,479 | -82% | 1 | 1 | 0% | 2,236 | 1,543 | -31% | 0 | 0 | — |
case-18 | fail→pass | 11,202 | 1,619 | -86% | 1 | 1 | 0% | 2,127 | 1,430 | -33% | 0 | 0 | — |
case-19 | fail→pass | 11,603 | 2,386 | -79% | 1 | 1 | 0% | 1,740 | 1,555 | -11% | 0 | 0 | — |
case-20 | pass→pass | 5,620 | 1,611 | -71% | 1 | 1 | 0% | 910 | 1,374 | +51% | 0 | 0 | — |
case-21 | pass→pass | 5,281 | 3,797 | -28% | 1 | 1 | 0% | 883 | 1,769 | +100% | 0 | 0 | — |
case-22 | pass→pass | 2,356 | 2,512 | +7% | 1 | 1 | 0% | 391 | 1,529 | +291% | 0 | 0 | — |
case-23 | pass→pass | 7,368 | 3,763 | -49% | 1 | 1 | 0% | 1,235 | 1,786 | +45% | 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 20 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 +52 percentage points is the difference between those two pass rates over the 20 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.