Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Systematically generate boundary inputs — boundary values, adversarial constructions, distribution shifts, rare combinations, scale extremes.
.claude/skills/yogsoth-ai-edge-case-generation/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-20 | ✓→✓ | ＝ Same ✓ | -32% | 0% |
| case-21 | ✓→✓ | ＝ Same ✓ | 9% | 0% |
| case-22 | ✓→✓ | ＝ Same ✓ | 43% | 0% |
| case-23 | ✓→✓ | ＝ Same ✓ | 19% | 0% |
| case-01 | ✗→✗ | ＝ Same ✗ | 29% | 0% |
Generate systematic edge cases to probe failures.
Subagent — spawned via subagent-spawning/spawn-agent.
One unit = one edge case generation pass (minimum 20 cases).
<!-- BEGIN available-tables (generated) -->
Optional, no fixed order; the final leaf is always a sop.
| SOP | When to use | | --- | --- | | spawn-agent | Spawn a customized CC subagent with full MCP tool access. Used by SOPs that declare execution: subagent. |
<!-- 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→fail | 21,805 | 26,643 | +22% | 1 | 1 | 0% | 3,378 | 4,358 | +29% | 0 | 0 | — |
case-02 | fail→fail | 24,202 | 41,322 | +71% | 1 | 1 | 0% | 3,679 | 6,705 | +82% | 0 | 0 | — |
case-03 | fail→fail | 12,586 | 31,757 | +152% | 1 | 1 | 0% | 1,966 | 5,399 | +175% | 0 | 0 | — |
case-04 | fail→fail | 9,737 | 12,323 | +27% | 1 | 1 | 0% | 1,544 | 2,185 | +42% | 0 | 0 | — |
case-05 | fail→fail | 18,893 | 22,094 | +17% | 1 | 1 | 0% | 2,810 | 3,417 | +22% | 0 | 0 | — |
case-06 | fail→fail | 24,159 | 25,725 | +6% | 1 | 1 | 0% | 4,013 | 4,063 | +1% | 0 | 0 | — |
case-07 | fail→fail | 18,271 | 23,860 | +31% | 1 | 1 | 0% | 2,827 | 3,679 | +30% | 0 | 0 | — |
case-08 | fail→fail | 10,514 | 9,667 | -8% | 1 | 1 | 0% | 1,610 | 1,630 | +1% | 0 | 0 | — |
case-09 | fail→fail | 20,921 | 27,661 | +32% | 1 | 1 | 0% | 3,233 | 4,300 | +33% | 0 | 0 | — |
case-10 | fail→fail | 17,874 | 20,781 | +16% | 1 | 1 | 0% | 2,070 | 2,530 | +22% | 0 | 0 | — |
case-11 | fail→fail | 22,465 | 34,908 | +55% | 1 | 1 | 0% | 3,496 | 5,189 | +48% | 0 | 0 | — |
case-12 | fail→fail | 20,220 | 32,243 | +59% | 1 | 1 | 0% | 3,169 | 4,998 | +58% | 0 | 0 | — |
case-13 | fail→fail | 22,464 | 26,845 | +20% | 1 | 1 | 0% | 3,282 | 4,507 | +37% | 0 | 0 | — |
case-14 | fail→fail | 23,642 | 26,600 | +13% | 1 | 1 | 0% | 3,659 | 4,144 | +13% | 0 | 0 | — |
case-15 | fail→fail | 21,430 | 34,526 | +61% | 1 | 1 | 0% | 3,375 | 5,394 | +60% | 0 | 0 | — |
case-16 | fail→fail | 18,147 | 26,169 | +44% | 1 | 1 | 0% | 2,825 | 3,933 | +39% | 0 | 0 | — |
case-17 | fail→fail | 20,392 | 19,379 | -5% | 1 | 1 | 0% | 2,858 | 2,881 | +1% | 0 | 0 | — |
case-18 | fail→fail | 21,500 | 37,903 | +76% | 1 | 1 | 0% | 3,307 | 5,184 | +57% | 0 | 0 | — |
case-19 | fail→fail | 17,523 | 24,217 | +38% | 1 | 1 | 0% | 2,680 | 2,974 | +11% | 0 | 0 | — |
case-20 | pass→pass | 12,236 | 7,469 | -39% | 1 | 1 | 0% | 2,630 | 1,786 | -32% | 0 | 0 | — |
case-21 | pass→pass | 12,103 | 12,032 | -1% | 1 | 1 | 0% | 2,000 | 2,180 | +9% | 0 | 0 | — |
case-22 | pass→pass | 17,888 | 25,336 | +42% | 1 | 1 | 0% | 2,899 | 4,151 | +43% | 0 | 0 | — |
case-23 | pass→pass | 15,242 | 16,181 | +6% | 1 | 1 | 0% | 2,540 | 3,019 | +19% | 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 0 percentage points is the difference between those two pass rates over the 23 comparable cases.
Other measured skills in the registry, with their headline benchmark lift.