Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Assess how removable a constraint is with effort estimate and dependency analysis.
.claude/skills/yogsoth-ai-removability-assessment/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-06 | ✗→✓ | ▲ Improved | 60% | 0% |
| case-09 | ✗→✓ | ▲ Improved | 61% | 0% |
| case-11 | ✗→✓ | ▲ Improved | 31% | 0% |
| case-12 | ✗→✓ | ▲ Improved | 29% | 0% |
| case-14 | ✗→✓ | ▲ Improved | 29% | 0% |
For a given constraint, assess how removable it is on a 0-1 scale, estimate the effort required to remove it, and identify dependencies that affect removability.
Spawns a subagent that:
Removability assessment requires focused analysis of a single constraint's characteristics, including research into analogous situations where similar constraints were or were not removed.
Output MUST include: removability score (0.0-1.0), effort estimate, and at least 1 dependency identified. Reject if score is provided without supporting rationale.
<!-- 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 | 20,692 | 21,710 | +5% | 1 | 1 | 0% | 3,118 | 3,035 | -3% | 0 | 0 | — |
case-02 | pass→fail | 18,523 | 33,442 | +81% | 1 | 1 | 0% | 3,745 | 5,979 | +60% | 0 | 0 | — |
case-03 | pass→pass | 17,174 | 18,132 | +6% | 1 | 1 | 0% | 2,790 | 3,305 | +18% | 0 | 0 | — |
case-04 | pass→pass | 48,639 | 34,515 | -29% | 1 | 1 | 0% | 4,022 | 4,977 | +24% | 0 | 0 | — |
case-05 | pass→pass | 15,523 | 12,066 | -22% | 1 | 1 | 0% | 1,440 | 1,347 | -6% | 0 | 0 | — |
case-06 | fail→pass | 18,776 | 24,415 | +30% | 1 | 1 | 0% | 2,105 | 3,361 | +60% | 0 | 0 | — |
case-07 | fail→fail | 26,614 | 26,093 | -2% | 1 | 1 | 0% | 3,271 | 3,384 | +3% | 0 | 0 | — |
case-08 | fail→fail | 23,371 | 24,945 | +7% | 1 | 1 | 0% | 2,741 | 3,422 | +25% | 0 | 0 | — |
case-09 | fail→pass | 17,839 | 22,688 | +27% | 1 | 1 | 0% | 1,971 | 3,165 | +61% | 0 | 0 | — |
case-10 | fail→fail | 28,453 | 31,472 | +11% | 1 | 1 | 0% | 3,513 | 4,298 | +22% | 0 | 0 | — |
case-11 | fail→pass | 24,452 | 27,581 | +13% | 1 | 1 | 0% | 2,983 | 3,899 | +31% | 0 | 0 | — |
case-12 | fail→pass | 21,848 | 32,747 | +50% | 1 | 1 | 0% | 2,512 | 3,230 | +29% | 0 | 0 | — |
case-13 | fail→fail | 21,404 | 25,349 | +18% | 1 | 1 | 0% | 2,542 | 2,114 | -17% | 0 | 0 | — |
case-14 | fail→pass | 24,184 | 22,862 | -5% | 1 | 1 | 0% | 3,042 | 3,917 | +29% | 0 | 0 | — |
case-15 | fail→fail | 22,224 | 19,315 | -13% | 1 | 1 | 0% | 2,574 | 3,400 | +32% | 0 | 0 | — |
case-16 | fail→pass | 23,008 | 21,640 | -6% | 1 | 1 | 0% | 2,748 | 2,802 | +2% | 0 | 0 | — |
case-17 | fail→pass | 22,525 | 23,730 | +5% | 1 | 1 | 0% | 2,499 | 3,296 | +32% | 0 | 0 | — |
case-18 | pass→pass | 22,340 | 34,609 | +55% | 1 | 1 | 0% | 2,734 | 3,276 | +20% | 0 | 0 | — |
case-19 | fail→fail | 24,116 | 24,149 | +0% | 1 | 1 | 0% | 2,791 | 3,414 | +22% | 0 | 0 | — |
case-20 | fail→pass | 8,431 | 20,397 | +142% | 1 | 1 | 0% | 487 | 2,749 | +464% | 0 | 0 | — |
case-21 | fail→pass | 21,669 | 31,573 | +46% | 1 | 1 | 0% | 3,155 | 3,593 | +14% | 0 | 0 | — |
case-22 | fail→pass | 22,413 | 20,375 | -9% | 1 | 1 | 0% | 2,478 | 3,599 | +45% | 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. The headline lift of +41 percentage points is the difference between those two pass rates over the 22 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.