Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Create phases to close all gaps identified by milestone audit
.claude/skills/davepoon-gsd-plan-milestone-gaps/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-21 | ✗→✓ | ▲ Improved | -42% | 0% |
| case-07 | ✗→✓ | ▲ Improved | 21% | 0% |
| case-08 | ✗→✓ | ▲ Improved | -52% | 0% |
| case-09 | ✗→✓ | ▲ Improved | -20% | 0% |
| case-13 | ✗→✓ | ▲ Improved | 0% | 0% |
<objective> Create all phases necessary to close gaps identified by /gsd:audit-milestone.
Reads MILESTONE-AUDIT.md, groups gaps into logical phases, creates phase entries in ROADMAP.md, and offers to plan each phase.
One command creates all fix phases — no manual /gsd:add-phase per gap. </objective>
<execution_context> @${CLAUDE_PLUGIN_ROOT}/workflows/plan-milestone-gaps.md </execution_context>
<context> Audit results: Glob: .planning/v-MILESTONE-AUDIT.md (use most recent)
Original intent and current planning state are loaded on demand inside the workflow. </context>
<process> Execute the plan-milestone-gaps workflow from @${CLAUDE_PLUGIN_ROOT}/workflows/plan-milestone-gaps.md end-to-end. Preserve all workflow gates (audit loading, prioritization, phase grouping, user confirmation, roadmap updates). </process>
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-21 | fail→pass | 11,071 | 5,136 | -54% | 1 | 1 | 0% | 1,747 | 1,016 | -42% | 0 | 0 | — |
case-01 | fail→fail | 3,770 | 3,959 | +5% | 1 | 1 | 0% | 624 | 446 | -29% | 0 | 0 | — |
case-02 | fail→fail | 12,249 | 3,454 | -72% | 1 | 1 | 0% | 1,827 | 413 | -77% | 0 | 0 | — |
case-03 | fail→fail | 6,220 | 4,474 | -28% | 1 | 1 | 0% | 1,000 | 414 | -59% | 0 | 0 | — |
case-04 | fail→fail | 6,022 | 5,814 | -3% | 1 | 1 | 0% | 987 | 540 | -45% | 0 | 0 | — |
case-05 | pass→pass | 4,887 | 3,536 | -28% | 1 | 1 | 0% | 834 | 845 | +1% | 0 | 0 | — |
case-06 | pass→pass | 10,922 | 6,364 | -42% | 1 | 1 | 0% | 1,992 | 1,361 | -32% | 0 | 0 | — |
case-07 | fail→pass | 5,909 | 5,474 | -7% | 1 | 1 | 0% | 1,090 | 1,319 | +21% | 0 | 0 | — |
case-08 | fail→pass | 10,277 | 3,770 | -63% | 1 | 1 | 0% | 1,644 | 788 | -52% | 0 | 0 | — |
case-09 | fail→pass | 9,696 | 6,510 | -33% | 1 | 1 | 0% | 1,595 | 1,269 | -20% | 0 | 0 | — |
case-10 | pass→pass | 6,166 | 1,691 | -73% | 1 | 1 | 0% | 1,124 | 524 | -53% | 0 | 0 | — |
case-11 | pass→pass | 4,565 | 2,787 | -39% | 1 | 1 | 0% | 750 | 733 | -2% | 0 | 0 | — |
case-12 | pass→pass | 8,131 | 4,686 | -42% | 1 | 1 | 0% | 1,470 | 998 | -32% | 0 | 0 | — |
case-13 | fail→pass | 4,645 | 3,751 | -19% | 1 | 1 | 0% | 805 | 805 | 0% | 0 | 0 | — |
case-14 | pass→pass | 11,173 | 3,999 | -64% | 1 | 1 | 0% | 1,799 | 830 | -54% | 0 | 0 | — |
case-15 | fail→pass | 10,149 | 2,020 | -80% | 1 | 1 | 0% | 1,718 | 576 | -66% | 0 | 0 | — |
case-16 | pass→pass | 9,408 | 2,663 | -72% | 1 | 1 | 0% | 1,584 | 724 | -54% | 0 | 0 | — |
case-17 | fail→pass | 10,648 | 1,240 | -88% | 1 | 1 | 0% | 1,835 | 422 | -77% | 0 | 0 | — |
case-18 | pass→fail | 8,283 | 4,657 | -44% | 1 | 1 | 0% | 1,353 | 912 | -33% | 0 | 0 | — |
case-19 | fail→pass | 7,859 | 3,397 | -57% | 1 | 1 | 0% | 1,222 | 826 | -32% | 0 | 0 | — |
case-20 | fail→pass | 7,741 | 2,407 | -69% | 1 | 1 | 0% | 1,237 | 656 | -47% | 0 | 0 | — |
case-22 | pass→pass | 8,554 | 2,951 | -66% | 1 | 1 | 0% | 1,368 | 710 | -48% | 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, and 18 counted toward the lift figure. The other 4 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 +36 percentage points is the difference between those two pass rates over the 18 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.