Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Legacy router from PRD/spec intent to the current Matt Pocock spec and ticket flow. Prefer to-spec and to-tickets.
.claude/skills/asymmetric-al-prd-to-plan/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-01 | ✗→✓ | ▲ Improved | 33% | 0% |
| case-02 | ✗→✓ | ▲ Improved | -66% | 0% |
| case-03 | ✗→✓ | ▲ Improved | 36% | 0% |
| case-04 | ✗→✓ | ▲ Improved | -69% | 0% |
| case-06 | ✗→✓ | ▲ Improved | -59% | 0% |
There is no prd-to-plan skill in mattpocock/skills. Use the current spec/ticket flow instead:
to-spec - docs/ai/skills/to-spec/SKILL.md - produce the spec(align with OpenSpec: openspec/specs/**, openspec/changes/**).
to-tickets - docs/ai/skills/to-tickets/SKILL.md - break the specinto tracer-bullet tickets with blocking edges.
implement, code-review, domain-modeling, andimprove-codebase-architecture as the work moves from tickets to code.
See references/upstream.md.
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→pass | 3,867 | 3,196 | -17% | 1 | 1 | 0% | 593 | 786 | +33% | 0 | 0 | — |
case-02 | fail→pass | 14,912 | 3,808 | -74% | 1 | 1 | 0% | 2,492 | 855 | -66% | 0 | 0 | — |
case-03 | fail→pass | 3,656 | 3,365 | -8% | 1 | 1 | 0% | 596 | 811 | +36% | 0 | 0 | — |
case-04 | fail→pass | 18,527 | 4,484 | -76% | 1 | 1 | 0% | 2,907 | 888 | -69% | 0 | 0 | — |
case-05 | pass→pass | 9,101 | 3,609 | -60% | 1 | 1 | 0% | 1,590 | 833 | -48% | 0 | 0 | — |
case-06 | fail→pass | 10,840 | 3,154 | -71% | 1 | 1 | 0% | 1,609 | 662 | -59% | 0 | 0 | — |
case-07 | fail→pass | 9,230 | 2,174 | -76% | 1 | 1 | 0% | 1,513 | 528 | -65% | 0 | 0 | — |
case-08 | fail→pass | 8,990 | 2,195 | -76% | 1 | 1 | 0% | 1,598 | 518 | -68% | 0 | 0 | — |
case-09 | pass→pass | 13,927 | 9,704 | -30% | 1 | 1 | 0% | 2,107 | 1,730 | -18% | 0 | 0 | — |
case-18 | fail→pass | 10,682 | 3,679 | -66% | 1 | 1 | 0% | 1,739 | 804 | -54% | 0 | 0 | — |
case-10 | pass→pass | 16,063 | 7,631 | -52% | 1 | 1 | 0% | 2,494 | 1,442 | -42% | 0 | 0 | — |
case-11 | pass→pass | 7,235 | 3,588 | -50% | 1 | 1 | 0% | 938 | 775 | -17% | 0 | 0 | — |
case-12 | pass→pass | 6,486 | 2,000 | -69% | 1 | 1 | 0% | 948 | 504 | -47% | 0 | 0 | — |
case-13 | pass→pass | 6,911 | 2,734 | -60% | 1 | 1 | 0% | 975 | 585 | -40% | 0 | 0 | — |
case-19 | fail→pass | 15,008 | 5,013 | -67% | 1 | 1 | 0% | 2,290 | 995 | -57% | 0 | 0 | — |
case-14 | fail→pass | 9,439 | 3,722 | -61% | 1 | 1 | 0% | 1,402 | 804 | -43% | 0 | 0 | — |
case-15 | fail→pass | 6,874 | 3,042 | -56% | 1 | 1 | 0% | 974 | 579 | -41% | 0 | 0 | — |
case-16 | fail→pass | 9,042 | 2,945 | -67% | 1 | 1 | 0% | 1,439 | 733 | -49% | 0 | 0 | — |
case-17 | fail→pass | 13,419 | 4,325 | -68% | 1 | 1 | 0% | 2,087 | 977 | -53% | 0 | 0 | — |
case-20 | fail→pass | 11,755 | 4,112 | -65% | 1 | 1 | 0% | 1,885 | 943 | -50% | 0 | 0 | — |
case-21 | fail→fail | 7,455 | 4,428 | -41% | 1 | 1 | 0% | 1,268 | 934 | -26% | 0 | 0 | — |
case-22 | fail→pass | 13,636 | 7,518 | -45% | 1 | 1 | 0% | 2,376 | 1,376 | -42% | 0 | 0 | — |
case-23 | pass→pass | 7,198 | 2,659 | -63% | 1 | 1 | 0% | 1,191 | 611 | -49% | 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 +65 percentage points is the difference between those two pass rates over the 23 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.