Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Map the complete validity envelope of a claim across all relevant dimensions, synthesizing breakpoints into a bounded region.
.claude/skills/yogsoth-ai-stress-test-validity-envelope-mapping/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-01 | ✗→✓ | ▲ Improved | 67% | 0% |
| case-03 | ✗→✓ | ▲ Improved | 32% | 0% |
| case-06 | ✗→✓ | ▲ Improved | 291% | 0% |
| case-07 | ✗→✓ | ▲ Improved | 6% | 0% |
| case-08 | ✗→✓ | ▲ Improved | -36% | 0% |
| Size | Envelope dimensions | Probes per dimension | Resolution | |---|---|---|---| | S | 2 | 5 | Coarse | | M | 4 | 8 | Medium | | L | 6 | 12 | Fine |
parameter-space-mapping for full dimension inventoryextreme-value-generationbreakpoint-detection with binary search refinementvalidity-envelope-construction for final synthesis<!-- BEGIN available-tables (generated) -->
Optional, no fixed order; the final leaf is always a sop.
| Tactic | When to use | | --- | --- | | boundary-probing | Map parameter space, generate extreme values, test at boundaries, detect breakpoints, synthesize validity envelope. | | contradiction-derivation | Negate a claim, derive logical consequences step by step, detect whether a genuine contradiction or absurdity emerges. |
Optional, no fixed order; the final leaf is always a sop.
| SOP | When to use | | --- | --- | | breakpoint-detection | Test a claim at extreme parameter values and detect the precise point where it breaks down. | | extreme-value-generation | Generate boundary and extreme test values for a given parameter dimension to stress-test claims. | | parameter-space-mapping | Identify all parameter dimensions along which a claim's validity might vary. | | stress-test-validity-envelope-construction | Synthesize breakpoints across dimensions into a coherent validity envelope for a claim. |
<!-- END available-tables (generated) -->
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→pass | 38,774 | 53,210 | +37% | 1 | 1 | 0% | 5,019 | 8,397 | +67% | 0 | 0 | — |
case-02 | fail→fail | 26,243 | 34,867 | +33% | 1 | 1 | 0% | 4,155 | 7,517 | +81% | 0 | 0 | — |
case-03 | fail→pass | 22,279 | 25,824 | +16% | 1 | 1 | 0% | 2,882 | 3,799 | +32% | 0 | 0 | — |
case-04 | pass→pass | 21,705 | 12,873 | -41% | 1 | 1 | 0% | 2,835 | 2,793 | -1% | 0 | 0 | — |
case-05 | pass→pass | 22,342 | 20,249 | -9% | 1 | 1 | 0% | 3,721 | 3,334 | -10% | 0 | 0 | — |
case-20 | pass→pass | 21,642 | 26,062 | +20% | 1 | 1 | 0% | 2,419 | 3,754 | +55% | 0 | 0 | — |
case-06 | fail→pass | 21,257 | 21,180 | -0% | 1 | 1 | 0% | 902 | 3,525 | +291% | 0 | 0 | — |
case-07 | fail→pass | 27,487 | 18,237 | -34% | 1 | 1 | 0% | 3,965 | 4,209 | +6% | 0 | 0 | — |
case-08 | fail→pass | 34,242 | 2,637 | -92% | 1 | 1 | 0% | 1,419 | 904 | -36% | 0 | 0 | — |
case-09 | fail→pass | 27,032 | 8,323 | -69% | 1 | 1 | 0% | 1,300 | 1,049 | -19% | 0 | 0 | — |
case-19 | pass→pass | 30,403 | 27,578 | -9% | 1 | 1 | 0% | 3,638 | 5,613 | +54% | 0 | 0 | — |
case-10 | fail→pass | 34,408 | 7,800 | -77% | 1 | 1 | 0% | 1,406 | 1,008 | -28% | 0 | 0 | — |
case-11 | pass→pass | 13,515 | 20,313 | +50% | 1 | 1 | 0% | 1,533 | 2,992 | +95% | 0 | 0 | — |
case-12 | pass→pass | 19,434 | 14,690 | -24% | 1 | 1 | 0% | 2,789 | 3,008 | +8% | 0 | 0 | — |
case-13 | fail→pass | 45,398 | 26,296 | -42% | 1 | 1 | 0% | 2,961 | 3,654 | +23% | 0 | 0 | — |
case-14 | pass→pass | 11,318 | 20,690 | +83% | 1 | 1 | 0% | 1,676 | 3,894 | +132% | 0 | 0 | — |
case-15 | pass→pass | 17,965 | 9,236 | -49% | 1 | 1 | 0% | 2,043 | 1,179 | -42% | 0 | 0 | — |
case-16 | pass→fail | 15,129 | 7,748 | -49% | 1 | 1 | 0% | 1,653 | 931 | -44% | 0 | 0 | — |
case-17 | pass→pass | 14,244 | 7,513 | -47% | 1 | 1 | 0% | 1,400 | 944 | -33% | 0 | 0 | — |
case-18 | fail→pass | 17,817 | 10,331 | -42% | 1 | 1 | 0% | 1,718 | 907 | -47% | 0 | 0 | — |
case-21 | pass→fail | 14,982 | 21,606 | +44% | 1 | 1 | 0% | 2,737 | 4,293 | +57% | 0 | 0 | — |
case-22 | pass→pass | 25,121 | 19,708 | -22% | 1 | 1 | 0% | 3,264 | 3,036 | -7% | 0 | 0 | — |
case-23 | pass→fail | 43,182 | 48,149 | +12% | 1 | 1 | 0% | 3,267 | 2,293 | -30% | 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, and 19 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 +26 percentage points is the difference between those two pass rates over the 19 comparable cases. 3 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.