Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Intent-driven issue lifecycle management — describe what you want in natural language (报告一个 bug / 列出开放 issue / 关掉 ISS-xxx / 关联到 task / 扫描发现问题) and the workflow routes to the right operation. Operates on .workflow/issues/. 知识管理走 /maestro-knowledge；knowhow 沉淀走 /maestro-knowhow；约束规则走 /maestro-spec。Triggers on "issue 管理", "报 bug", "记录问题", "issue list", "关闭 issue", "issue discover", "发现问题".
.claude/skills/catlog22-maestro-issue/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-01 | ✗→✓ | ▲ Improved | 113% | 0% |
| case-02 | ✗→✓ | ▲ Improved | -10% | 0% |
| case-04 | ✗→✓ | ▲ Improved | 790% | 0% |
| case-23 | ✗→✓ | ▲ Improved | 195% | 0% |
| case-07 | ✗→✓ | ▲ Improved | 8% | 0% |
<purpose> Intent-driven issue management (renamed from maestro-manage, narrowed to issues). No fixed subcommand grammar — state your intent; the issue step classifies it into one operation and extracts the needed parameters:
</purpose>
<dispatch> Execute the issue step inside a v3 Session: open one with maestro session open "<objective>" --id <slug> --chain issue --participant {p} --actor {a} --request-id {r} --reason "<reason>" --json and dispatch with fenced maestro run next --session {session_id} ... --expected-orchestration-revision {rev} --json (or self-start with maestro run create issue [args...] --session {session_id} ... --json), passing the full $ARGUMENTS as the step input (repeatable --arg / positional passthrough per the command contract). Read context read-only with maestro session status / maestro session resume-view; maestro run complete {run_id} ... --advance publishes outputs and auto-stages knowledge candidates. (v2's run skill dispatcher is removed from the v3 surface.)
The step classifies the intent, extracts parameters, and routes to the operation.
create|list|status|show|update|close|link) and --flags still work as deterministic shortcuts and override inferred values.discover routes to the dedicated issue-discover step.</dispatch>
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→pass | 4,062 | 4,744 | +17% | 1 | 1 | 0% | 575 | 1,223 | +113% | 0 | 0 | — |
case-02 | fail→pass | 6,201 | 2,595 | -58% | 1 | 1 | 0% | 879 | 793 | -10% | 0 | 0 | — |
case-03 | fail→fail | 8,434 | 9,973 | +18% | 1 | 1 | 0% | 1,140 | 1,047 | -8% | 0 | 0 | — |
case-04 | fail→pass | 2,763 | 16,692 | +504% | 1 | 1 | 0% | 356 | 3,170 | +790% | 0 | 0 | — |
case-23 | fail→pass | 4,610 | 7,741 | +68% | 1 | 1 | 0% | 598 | 1,763 | +195% | 0 | 0 | — |
case-05 | fail→fail | 4,169 | 4,899 | +18% | 1 | 1 | 0% | 734 | 1,090 | +49% | 0 | 0 | — |
case-06 | fail→fail | 8,332 | 14,407 | +73% | 1 | 1 | 0% | 1,383 | 2,720 | +97% | 0 | 0 | — |
case-07 | fail→pass | 6,959 | 3,926 | -44% | 1 | 1 | 0% | 946 | 1,024 | +8% | 0 | 0 | — |
case-08 | pass→fail | 7,421 | 6,939 | -6% | 1 | 1 | 0% | 957 | 1,683 | +76% | 0 | 0 | — |
case-09 | pass→pass | 10,716 | 3,412 | -68% | 1 | 1 | 0% | 1,673 | 936 | -44% | 0 | 0 | — |
case-10 | fail→pass | 12,517 | 2,604 | -79% | 1 | 1 | 0% | 1,734 | 807 | -53% | 0 | 0 | — |
case-11 | pass→fail | 3,005 | 9,613 | +220% | 1 | 1 | 0% | 333 | 2,041 | +513% | 0 | 0 | — |
case-12 | fail→pass | 9,794 | 2,821 | -71% | 1 | 1 | 0% | 1,527 | 846 | -45% | 0 | 0 | — |
case-13 | fail→pass | 3,024 | 6,827 | +126% | 1 | 1 | 0% | 368 | 1,583 | +330% | 0 | 0 | — |
case-14 | pass→fail | 3,494 | 7,885 | +126% | 1 | 1 | 0% | 433 | 827 | +91% | 0 | 0 | — |
case-15 | fail→fail | 2,884 | 3,160 | +10% | 1 | 1 | 0% | 361 | 931 | +158% | 0 | 0 | — |
case-16 | fail→pass | 5,787 | 9,536 | +65% | 1 | 1 | 0% | 835 | 1,410 | +69% | 0 | 0 | — |
case-17 | fail→pass | 4,617 | 2,022 | -56% | 1 | 1 | 0% | 635 | 764 | +20% | 0 | 0 | — |
case-18 | fail→pass | 8,703 | 2,025 | -77% | 1 | 1 | 0% | 1,296 | 734 | -43% | 0 | 0 | — |
case-19 | fail→pass | 8,600 | 7,267 | -16% | 1 | 1 | 0% | 1,401 | 1,394 | -0% | 0 | 0 | — |
case-20 | pass→fail | 9,208 | 7,688 | -17% | 1 | 1 | 0% | 1,367 | 859 | -37% | 0 | 0 | — |
case-21 | pass→pass | 7,118 | 4,184 | -41% | 1 | 1 | 0% | 1,187 | 1,198 | +1% | 0 | 0 | — |
case-22 | fail→pass | 5,411 | 3,658 | -32% | 1 | 1 | 0% | 799 | 1,084 | +36% | 0 | 0 | — |
case-24 | pass→fail | 20,476 | 6,923 | -66% | 1 | 1 | 0% | 3,153 | 859 | -73% | 0 | 0 | — |
case-25 | fail→fail | 12,444 | 4,245 | -66% | 1 | 1 | 0% | 1,935 | 1,069 | -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. 25 cases were attempted, and 22 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 +32 percentage points is the difference between those two pass rates over the 22 comparable cases. 5 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.