Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Strategy: Design prevention, detection, and response measures for high-priority failure modes. Produces actionable countermeasures validated via re-scoring.
.claude/skills/yogsoth-ai-mitigation-design/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-01 | ✗→✓ | ▲ Improved | -9% | 0% |
| case-03 | ✗→✓ | ▲ Improved | 38% | 0% |
| case-06 | ✗→✓ | ▲ Improved | 32% | 0% |
| case-09 | ✗→✓ | ▲ Improved | -7% | 0% |
| case-16 | ✗→✓ | ▲ Improved | -48% | 0% |
Design countermeasures for high-priority failure modes across three layers: prevention, detection, response.
| Parameter | S | M | L | |---|---|---|---| | Mitigations per failure mode | 1 | 2 | 3 | | Re-scoring rounds | 1 | 2 | 3 | | Mitigation validation depth | shallow | standard | full FMEA |
[H-priority failures] → mitigation-design-sop
→ re-scoring → action-priority-matrix (re-classify)
→ [still H?] → mitigation-validation → iterate
→ [M or L?] → accept, document residual risk<!-- BEGIN available-tables (generated) -->
Optional, no fixed order; the final leaf is always a sop.
| Tactic | When to use | | --- | --- | | mitigation-validation | Tactic: Run mini-FMEA on proposed mitigations to verify they do not introduce new failure modes. Prevents mitigation-induced risks. |
Optional, no fixed order; the final leaf is always a sop.
| SOP | When to use | | --- | --- | | action-priority-matrix | Compute Risk Priority Number (RPN = S x O x D), classify failure modes into H/M/L action priority per AIAG-VDA tables. | | mitigation-design-sop | Design prevention, detection, and response measures for high-priority failure modes. Produces actionable countermeasure specifications. | | re-scoring | Re-evaluate S/O/D scores after mitigation measures are in place. Validates that mitigations actually reduce risk as expected. |
<!-- 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 | 37,533 | 32,021 | -15% | 1 | 1 | 0% | 6,277 | 5,716 | -9% | 0 | 0 | — |
case-02 | pass→pass | 26,549 | 23,903 | -10% | 1 | 1 | 0% | 4,407 | 4,194 | -5% | 0 | 0 | — |
case-03 | fail→pass | 32,037 | 25,253 | -21% | 1 | 1 | 0% | 2,890 | 3,988 | +38% | 0 | 0 | — |
case-04 | pass→pass | 16,860 | 18,625 | +10% | 1 | 1 | 0% | 2,633 | 2,683 | +2% | 0 | 0 | — |
case-05 | pass→pass | 15,281 | 17,517 | +15% | 1 | 1 | 0% | 1,589 | 2,724 | +71% | 0 | 0 | — |
case-06 | fail→pass | 18,854 | 19,462 | +3% | 1 | 1 | 0% | 2,919 | 3,843 | +32% | 0 | 0 | — |
case-07 | pass→pass | 53,327 | 30,807 | -42% | 1 | 1 | 0% | 4,075 | 5,050 | +24% | 0 | 0 | — |
case-08 | pass→pass | 11,886 | 12,990 | +9% | 1 | 1 | 0% | 1,819 | 1,831 | +1% | 0 | 0 | — |
case-09 | fail→pass | 16,875 | 12,156 | -28% | 1 | 1 | 0% | 1,786 | 1,655 | -7% | 0 | 0 | — |
case-10 | pass→pass | 21,489 | 15,390 | -28% | 1 | 1 | 0% | 2,696 | 2,404 | -11% | 0 | 0 | — |
case-11 | pass→pass | 18,608 | 14,493 | -22% | 1 | 1 | 0% | 1,982 | 1,977 | -0% | 0 | 0 | — |
case-12 | pass→pass | 14,719 | 10,959 | -26% | 1 | 1 | 0% | 2,179 | 2,022 | -7% | 0 | 0 | — |
case-13 | pass→pass | 18,567 | 15,194 | -18% | 1 | 1 | 0% | 2,080 | 2,050 | -1% | 0 | 0 | — |
case-14 | pass→pass | 20,632 | 15,904 | -23% | 1 | 1 | 0% | 2,698 | 2,356 | -13% | 0 | 0 | — |
case-15 | pass→pass | 13,249 | 26,242 | +98% | 1 | 1 | 0% | 2,021 | 2,773 | +37% | 0 | 0 | — |
case-16 | fail→pass | 16,836 | 7,489 | -56% | 1 | 1 | 0% | 1,769 | 918 | -48% | 0 | 0 | — |
case-17 | pass→pass | 20,490 | 7,416 | -64% | 1 | 1 | 0% | 2,396 | 941 | -61% | 0 | 0 | — |
case-18 | fail→pass | 19,423 | 1,914 | -90% | 1 | 1 | 0% | 2,102 | 856 | -59% | 0 | 0 | — |
case-19 | pass→pass | 9,619 | 10,959 | +14% | 1 | 1 | 0% | 1,461 | 1,557 | +7% | 0 | 0 | — |
case-20 | pass→fail | 22,604 | 40,352 | +79% | 1 | 1 | 0% | 2,936 | 6,390 | +118% | 0 | 0 | — |
case-21 | fail→fail | 14,517 | 39,770 | +174% | 1 | 1 | 0% | 2,202 | 6,065 | +175% | 0 | 0 | — |
case-22 | pass→pass | 16,616 | 22,737 | +37% | 1 | 1 | 0% | 2,728 | 3,190 | +17% | 0 | 0 | — |
case-23 | pass→pass | 9,975 | 8,153 | -18% | 1 | 1 | 0% | 1,666 | 1,923 | +15% | 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 +22 percentage points is the difference between those two pass rates over the 23 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.