Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →You are a GitHub issue resolution expert specializing in systematic bug investigation, feature implementation, and collaborative development workflows. Your expertise spans issue triage, root cause an
.claude/skills/dokhacgiakhoa-team-collaboration-issue/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-03 | ✗→✓ | ▲ Improved | -15% | 0% |
| case-09 | ✓→✓ | ＝ Same ✓ | 48% | 0% |
| case-02 | ✓→✓ | ＝ Same ✓ | 10% | 0% |
| case-07 | ✓→✓ | ＝ Same ✓ | -6% | 0% |
| case-08 | ✓→✓ | ＝ Same ✓ | 46% | 0% |
You are a GitHub issue resolution expert specializing in systematic bug investigation, feature implementation, and collaborative development workflows. Your expertise spans issue triage, root cause analysis, test-driven development, and pull request management. You excel at transforming vague bug reports into actionable fixes and feature requests into production-ready code.
The user needs comprehensive GitHub issue resolution that goes beyond simple fixes. Focus on thorough investigation, proper branch management, systematic implementation with testing, and professional pull request creation that follows modern CI/CD practices.
GitHub Issue ID or URL: $ARGUMENTS
resources/implementation-playbook.md.resources/implementation-playbook.md for detailed patterns and examples.| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-09 | pass→pass | 12,717 | 24,576 | +93% | 1 | 1 | 0% | 2,060 | 3,053 | +48% | 0 | 0 | — |
case-01 | fail→fail | 25,034 | 20,599 | -18% | 1 | 1 | 0% | 3,731 | 2,844 | -24% | 0 | 0 | — |
case-02 | pass→pass | 22,075 | 22,122 | +0% | 1 | 1 | 0% | 2,367 | 2,606 | +10% | 0 | 0 | — |
case-03 | fail→pass | 22,194 | 18,466 | -17% | 1 | 1 | 0% | 3,087 | 2,629 | -15% | 0 | 0 | — |
case-04 | fail→fail | 18,460 | 15,510 | -16% | 1 | 1 | 0% | 2,070 | 2,672 | +29% | 0 | 0 | — |
case-05 | fail→fail | 15,687 | 27,430 | +75% | 1 | 1 | 0% | 2,438 | 3,389 | +39% | 0 | 0 | — |
case-06 | fail→fail | 11,317 | 16,031 | +42% | 1 | 1 | 0% | 1,915 | 2,189 | +14% | 0 | 0 | — |
case-07 | pass→pass | 20,910 | 16,978 | -19% | 1 | 1 | 0% | 2,685 | 2,535 | -6% | 0 | 0 | — |
case-08 | pass→pass | 17,347 | 20,712 | +19% | 1 | 1 | 0% | 2,070 | 3,019 | +46% | 0 | 0 | — |
case-10 | fail→fail | 18,449 | 30,197 | +64% | 1 | 1 | 0% | 2,855 | 3,999 | +40% | 0 | 0 | — |
case-11 | fail→fail | 19,544 | 22,189 | +14% | 1 | 1 | 0% | 2,033 | 2,911 | +43% | 0 | 0 | — |
case-12 | fail→fail | 18,602 | 18,394 | -1% | 1 | 1 | 0% | 2,756 | 3,587 | +30% | 0 | 0 | — |
case-13 | fail→fail | 17,230 | 22,003 | +28% | 1 | 1 | 0% | 2,839 | 2,738 | -4% | 0 | 0 | — |
case-14 | fail→fail | 18,826 | 13,017 | -31% | 1 | 1 | 0% | 2,397 | 2,607 | +9% | 0 | 0 | — |
case-15 | pass→pass | 19,352 | 19,042 | -2% | 1 | 1 | 0% | 2,135 | 3,168 | +48% | 0 | 0 | — |
case-16 | fail→fail | 15,078 | 21,463 | +42% | 1 | 1 | 0% | 2,409 | 3,089 | +28% | 0 | 0 | — |
case-17 | fail→fail | 25,659 | 19,718 | -23% | 1 | 1 | 0% | 3,501 | 3,942 | +13% | 0 | 0 | — |
case-18 | pass→pass | 18,723 | 15,516 | -17% | 1 | 1 | 0% | 2,267 | 2,911 | +28% | 0 | 0 | — |
case-19 | pass→pass | 7,876 | 8,136 | +3% | 1 | 1 | 0% | 410 | 820 | +100% | 0 | 0 | — |
case-20 | pass→pass | 8,461 | 19,953 | +136% | 1 | 1 | 0% | 1,571 | 1,597 | +2% | 0 | 0 | — |
case-21 | pass→pass | 10,877 | 13,240 | +22% | 1 | 1 | 0% | 1,108 | 1,848 | +67% | 0 | 0 | — |
case-22 | pass→pass | 16,349 | 10,547 | -35% | 1 | 1 | 0% | 1,766 | 2,147 | +22% | 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. The headline lift of +5 percentage points is the difference between those two pass rates over the 22 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.