Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →SOP for creating a dimension page — documents an axis of variation with its values and semantics.
.claude/skills/yogsoth-ai-dimension-page-creation/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-04 | ✗→✓ | ▲ Improved | 57% | 0% |
| case-17 | ✗→✓ | ▲ Improved | 202% | 0% |
| case-12 | ✓→✗ | ▼ Worse | -70% | 0% |
| case-08 | ✓→✗ | ▼ Worse | 128% | 0% |
| case-09 | ✓→✗ | ▼ Worse | -76% | 0% |
Create a page representing a dimension (axis of variation) in the design space.
CC file write + vault_add_edge + vault_index
concepts/<dimension-slug>.md with frontmatter (tags: dimension])[[dir/slug]] pointing to the target (dir/slug = target path minus .md). Place inline at semantically relevant location. Skip if already present.<HARD-GATE> Dimension must have ≥2 distinct values and a clear definition of what varies along the axis. </HARD-GATE>
Returns: { path: string, values: string[], edges_added: number }
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-12 | pass→fail | 10,339 | 4,557 | -56% | 1 | 1 | 0% | 1,634 | 492 | -70% | 0 | 0 | — |
case-13 | fail→fail | 15,369 | 6,598 | -57% | 1 | 1 | 0% | 745 | 719 | -3% | 0 | 0 | — |
case-06 | fail→fail | 4,081 | 4,223 | +3% | 1 | 1 | 0% | 196 | 567 | +189% | 0 | 0 | — |
case-01 | fail→fail | 4,232 | 5,281 | +25% | 1 | 1 | 0% | 265 | 538 | +103% | 0 | 0 | — |
case-02 | fail→fail | 15,079 | 5,259 | -65% | 1 | 1 | 0% | 2,488 | 571 | -77% | 0 | 0 | — |
case-03 | fail→fail | 14,000 | 4,379 | -69% | 1 | 1 | 0% | 2,422 | 508 | -79% | 0 | 0 | — |
case-04 | fail→pass | 5,691 | 7,329 | +29% | 1 | 1 | 0% | 944 | 1,480 | +57% | 0 | 0 | — |
case-05 | fail→fail | 7,427 | 3,962 | -47% | 1 | 1 | 0% | 895 | 511 | -43% | 0 | 0 | — |
case-07 | fail→fail | 4,938 | 4,662 | -6% | 1 | 1 | 0% | 265 | 520 | +96% | 0 | 0 | — |
case-08 | pass→fail | 2,189 | 4,411 | +102% | 1 | 1 | 0% | 208 | 474 | +128% | 0 | 0 | — |
case-09 | pass→fail | 12,697 | 5,186 | -59% | 1 | 1 | 0% | 2,168 | 512 | -76% | 0 | 0 | — |
case-10 | fail→fail | 7,395 | 6,384 | -14% | 1 | 1 | 0% | 1,406 | 635 | -55% | 0 | 0 | — |
case-11 | fail→fail | 22,821 | 4,948 | -78% | 1 | 1 | 0% | 3,758 | 502 | -87% | 0 | 0 | — |
case-14 | fail→fail | 9,607 | 4,762 | -50% | 1 | 1 | 0% | 1,632 | 606 | -63% | 0 | 0 | — |
case-15 | fail→fail | 18,304 | 4,602 | -75% | 1 | 1 | 0% | 3,020 | 572 | -81% | 0 | 0 | — |
case-16 | fail→fail | 7,834 | 5,451 | -30% | 1 | 1 | 0% | 1,355 | 589 | -57% | 0 | 0 | — |
case-17 | fail→pass | 3,916 | 8,446 | +116% | 1 | 1 | 0% | 579 | 1,747 | +202% | 0 | 0 | — |
case-18 | pass→fail | 10,478 | 5,228 | -50% | 1 | 1 | 0% | 1,823 | 540 | -70% | 0 | 0 | — |
case-19 | fail→fail | 9,111 | 4,911 | -46% | 1 | 1 | 0% | 1,529 | 556 | -64% | 0 | 0 | — |
case-20 | fail→fail | 16,153 | 5,722 | -65% | 1 | 1 | 0% | 2,963 | 564 | -81% | 0 | 0 | — |
case-21 | fail→fail | 7,905 | 5,601 | -29% | 1 | 1 | 0% | 1,320 | 650 | -51% | 0 | 0 | — |
case-22 | fail→fail | 23,148 | 5,549 | -76% | 1 | 1 | 0% | 4,571 | 587 | -87% | 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 2 counted toward the lift figure. The other 20 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 -9 percentage points is the difference between those two pass rates over the 2 comparable cases. 6 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.