Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Manage GitHub issues, PRs, repos, actions, and code search from the terminal. Use when an agent needs to create issues, open pull requests, manage releases, or search code on GitHub.
.claude/skills/composio-community-github-cli-gh/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-09 | ✗→✓ | ▲ Improved | 27% | 0% |
| case-07 | ✓→✓ | = Same ✓ | -55% | 0% |
| case-01 | ✓→✓ | = Same ✓ | -42% | 0% |
| case-02 | ✓→✓ | = Same ✓ | -50% | 0% |
| case-03 | ✓→✓ | = Same ✓ | -54% | 0% |
Issues, PRs, repos, actions, and code search from the terminal.
bashbrew install gh gh auth login
bashgh issue list --json number,title,state gh issue create --title "Bug" --body "Description" gh pr list --json number,title,headRefName gh pr create --title "Feature" --body "Description" gh pr merge <number> --squash gh repo clone owner/repo gh repo create my-repo --public gh run list --json status,name,conclusion gh api repos/owner/repo/releases --jq '.[0].tag_name' gh search code "function_name" --json path,repository
--json with --jq for precise field extractiongh api for arbitrary GitHub API callsGH_TOKEN or GITHUB_TOKEN| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-07 | pass→pass | 8,892 | 2,620 | -71% | 1 | 1 | 0% | 1,250 | 563 | -55% | 0 | 0 | — |
case-01 | pass→pass | 7,046 | 3,737 | -47% | 1 | 1 | 0% | 1,056 | 617 | -42% | 0 | 0 | — |
case-02 | pass→pass | 10,208 | 2,967 | -71% | 1 | 1 | 0% | 1,550 | 774 | -50% | 0 | 0 | — |
case-03 | pass→pass | 9,649 | 4,924 | -49% | 1 | 1 | 0% | 1,783 | 822 | -54% | 0 | 0 | — |
case-04 | pass→pass | 8,397 | 3,097 | -63% | 1 | 1 | 0% | 1,435 | 645 | -55% | 0 | 0 | — |
case-05 | pass→pass | 9,039 | 2,129 | -76% | 1 | 1 | 0% | 1,291 | 586 | -55% | 0 | 0 | — |
case-06 | pass→pass | 4,611 | 2,087 | -55% | 1 | 1 | 0% | 400 | 419 | +5% | 0 | 0 | — |
case-08 | pass→pass | 7,476 | 2,368 | -68% | 1 | 1 | 0% | 923 | 575 | -38% | 0 | 0 | — |
case-09 | fail→pass | 5,641 | 4,205 | -25% | 1 | 1 | 0% | 753 | 959 | +27% | 0 | 0 | — |
case-10 | pass→pass | 7,725 | 2,215 | -71% | 1 | 1 | 0% | 1,062 | 539 | -49% | 0 | 0 | — |
case-11 | pass→pass | 11,603 | 9,180 | -21% | 1 | 1 | 0% | 1,933 | 1,895 | -2% | 0 | 0 | — |
case-16 | pass→pass | 6,195 | 3,634 | -41% | 1 | 1 | 0% | 625 | 727 | +16% | 0 | 0 | — |
case-12 | pass→pass | 8,569 | 3,850 | -55% | 1 | 1 | 0% | 990 | 836 | -16% | 0 | 0 | — |
case-13 | pass→pass | 9,161 | 6,509 | -29% | 1 | 1 | 0% | 1,618 | 1,413 | -13% | 0 | 0 | — |
case-14 | pass→pass | 7,541 | 4,769 | -37% | 1 | 1 | 0% | 1,185 | 822 | -31% | 0 | 0 | — |
case-15 | pass→pass | 3,211 | 2,616 | -19% | 1 | 1 | 0% | 448 | 433 | -3% | 0 | 0 | — |
case-17 | pass→pass | 13,805 | 6,711 | -51% | 1 | 1 | 0% | 1,854 | 1,297 | -30% | 0 | 0 | — |
case-18 | pass→pass | 9,548 | 2,563 | -73% | 1 | 1 | 0% | 1,361 | 673 | -51% | 0 | 0 | — |
case-19 | pass→pass | 11,930 | 7,635 | -36% | 1 | 1 | 0% | 1,747 | 1,130 | -35% | 0 | 0 | — |
case-20 | pass→pass | 1,580 | 2,162 | +37% | 1 | 1 | 0% | 177 | 494 | +179% | 0 | 0 | — |
case-21 | pass→pass | 8,334 | 4,276 | -49% | 1 | 1 | 0% | 1,256 | 960 | -24% | 0 | 0 | — |
case-22 | pass→pass | 2,862 | 2,309 | -19% | 1 | 1 | 0% | 413 | 593 | +44% | 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.
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.