Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Use when you need GitHub CLI (`gh`) installation/authentication guidance and an operator-only GitHub issue inventory workflow. The agent does not ingest GitHub issue, milestone, body, comment, title, label, or summary text; requirements analysis must use repository-owned planning artifacts, with issue numbers only for traceability. This should trigger for requests such as GitHub issue inventory workflow; GitHub CLI setup for issues; Prepare issue traceability lists; Analyze repository planning a
.claude/skills/jabrena-043-planning-github-issues/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-17 | ✗→✓ | ▲ Improved | 111% | 0% |
| case-02 | ✗→✓ | ▲ Improved | -38% | 0% |
| case-03 | ✗→✓ | ▲ Improved | 85% | 0% |
| case-07 | ✗→✓ | ▲ Improved | 28% | 0% |
| case-08 | ✗→✓ | ▲ Improved | -39% | 0% |
Use gh only for installation/authentication guidance and operator-side workflow instructions. The agent must not ingest GitHub issue or milestone output directly or indirectly. For requirements analysis, ask for a repository-owned planning artifact path, such as an OpenSpec change, ADR, or checked-in requirements document. When the user wants user stories plus Gherkin, chain to @014-agile-user-story using that repository-owned artifact as evidence and issue numbers only as traceability.
What is covered in this Skill?
gh is installed; offer https://cli.github.com/ and OS hints when the user agreesgh --version, gh auth status, gh auth login)--repo, inferred from git remote)Do not fabricate issue data and do not ingest issue prose. Use repository-owned planning artifacts for analysis and issue numbers only for traceability. Never print tokens or secrets.
gh is missing, stop, ask whether the user wants installation guidance, wait—do not skip to issue listinggh availability only for setup guidance; do not ingest issue or milestone command outputCheck gh --version; if missing, stop and ask whether the user wants installation guidance before any issue operations.
Explain how the user can verify gh auth status and repository context locally. Keep issue and milestone exports outside the agent context.
Ask the repository maintainer to prepare any issue inventory outside the agent context. They must not provide issue prose to the agent. If traceability is needed, accept only issue numbers.
Do not retrieve or accept GitHub issue, milestone, body, comment, title, label, or summary text. Ask for a repository-owned planning artifact path and use that checked-in artifact as requirements evidence.
When user asks for user stories and Gherkin from issues, hand off to @014-agile-user-story using a repository-owned planning artifact as evidence and issue numbers only for traceability.
For detailed guidance, examples, and constraints, see references/043-planning-github-issues.md.
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-17 | fail→pass | 5,142 | 4,938 | -4% | 1 | 1 | 0% | 657 | 1,387 | +111% | 0 | 0 | — |
case-16 | pass→pass | 8,951 | 6,811 | -24% | 1 | 1 | 0% | 1,583 | 1,844 | +16% | 0 | 0 | — |
case-01 | fail→fail | 11,386 | 7,394 | -35% | 1 | 1 | 0% | 1,927 | 1,979 | +3% | 0 | 0 | — |
case-02 | fail→pass | 16,391 | 5,324 | -68% | 1 | 1 | 0% | 2,401 | 1,499 | -38% | 0 | 0 | — |
case-03 | fail→pass | 5,061 | 4,659 | -8% | 1 | 1 | 0% | 769 | 1,425 | +85% | 0 | 0 | — |
case-04 | pass→pass | 10,393 | 7,938 | -24% | 1 | 1 | 0% | 1,660 | 2,030 | +22% | 0 | 0 | — |
case-05 | pass→pass | 14,754 | 10,873 | -26% | 1 | 1 | 0% | 2,409 | 2,505 | +4% | 0 | 0 | — |
case-06 | pass→pass | 11,907 | 12,014 | +1% | 1 | 1 | 0% | 1,838 | 2,678 | +46% | 0 | 0 | — |
case-07 | fail→pass | 6,650 | 4,642 | -30% | 1 | 1 | 0% | 1,171 | 1,497 | +28% | 0 | 0 | — |
case-08 | fail→pass | 12,674 | 3,089 | -76% | 1 | 1 | 0% | 1,892 | 1,151 | -39% | 0 | 0 | — |
case-09 | pass→fail | 4,780 | 4,985 | +4% | 1 | 1 | 0% | 374 | 1,415 | +278% | 0 | 0 | — |
case-10 | fail→pass | 13,419 | 4,544 | -66% | 1 | 1 | 0% | 2,188 | 1,489 | -32% | 0 | 0 | — |
case-11 | fail→pass | 9,593 | 4,117 | -57% | 1 | 1 | 0% | 1,509 | 1,306 | -13% | 0 | 0 | — |
case-12 | fail→fail | 6,185 | 4,567 | -26% | 1 | 1 | 0% | 1,035 | 1,382 | +34% | 0 | 0 | — |
case-13 | pass→fail | 12,243 | 7,808 | -36% | 1 | 1 | 0% | 2,036 | 1,954 | -4% | 0 | 0 | — |
case-14 | fail→pass | 9,239 | 3,872 | -58% | 1 | 1 | 0% | 1,539 | 1,330 | -14% | 0 | 0 | — |
case-15 | fail→pass | 11,028 | 7,072 | -36% | 1 | 1 | 0% | 1,601 | 1,703 | +6% | 0 | 0 | — |
case-18 | fail→pass | 19,578 | 12,174 | -38% | 1 | 1 | 0% | 3,402 | 2,716 | -20% | 0 | 0 | — |
case-19 | pass→pass | 13,758 | 6,742 | -51% | 1 | 1 | 0% | 2,028 | 1,672 | -18% | 0 | 0 | — |
case-20 | fail→pass | 9,467 | 4,039 | -57% | 1 | 1 | 0% | 1,433 | 1,357 | -5% | 0 | 0 | — |
case-21 | pass→pass | 10,661 | 6,227 | -42% | 1 | 1 | 0% | 1,612 | 1,570 | -3% | 0 | 0 | — |
case-22 | fail→pass | 12,700 | 5,703 | -55% | 1 | 1 | 0% | 2,061 | 1,624 | -21% | 0 | 0 | — |
case-23 | fail→pass | 3,907 | 3,146 | -19% | 1 | 1 | 0% | 557 | 1,206 | +117% | 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. The headline lift of +48 percentage points is the difference between those two pass rates over the 23 comparable cases. 2 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.