Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Inject constraints → force creative response → extract transferable principles. Orchestrates constraint injection, response generation, and principle extraction.
.claude/skills/yogsoth-ai-constraint-protocol/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-02 | ✗→✓ | ▲ Improved | 52% | 0% |
| case-03 | ✗→✓ | ▲ Improved | 27% | 0% |
| case-05 | ✗→✓ | ▲ Improved | 32% | 0% |
| case-09 | ✗→✓ | ▲ Improved | -19% | 0% |
| case-11 | ✗→✓ | ▲ Improved | 5% | 0% |
Inject constraints → force creative response → extract transferable principles.
Generate extreme constraints across multiple types (resource, time, scale, audience, material, inversion, context). Each constraint should make the current approach impossible, forcing genuinely new thinking.
For each constraint, generate a creative solution that satisfies both the original problem AND the constraint. No "impossible" — find a way. Use constraint-response SOP.
From each constrained solution, extract the underlying principle that made it work. Identify which principles apply even WITHOUT the constraint — these are the creative gold.
| Metric | Floor | |--------|-------| | Constraints injected | ≥5 | | Constraint-forced solutions | ≥3 | | Transferable principles extracted | ≥3 | | Principles validated as novel | ≥2 |
| SOP | Role | |-----|------| | constraint-response | Stage 2 — generate solutions under constraints | | perspective-synthesis | Stage 3 — synthesize principles into report | | novelty-scoring | Post — score novelty of extracted principles |
<!-- BEGIN available-tables (generated) -->
Optional, no fixed order; the final leaf is always a sop.
| SOP | When to use | | --- | --- | | constraint-response | Generate creative solutions under extreme constraints — no "impossible" allowed, find a way. |
<!-- 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 | pass→pass | 25,109 | 32,188 | +28% | 1 | 1 | 0% | 3,570 | 4,812 | +35% | 0 | 0 | — |
case-02 | fail→pass | 21,371 | 27,560 | +29% | 1 | 1 | 0% | 2,881 | 4,383 | +52% | 0 | 0 | — |
case-03 | fail→pass | 19,989 | 28,137 | +41% | 1 | 1 | 0% | 3,664 | 4,637 | +27% | 0 | 0 | — |
case-04 | fail→fail | 16,313 | 21,866 | +34% | 1 | 1 | 0% | 2,230 | 3,295 | +48% | 0 | 0 | — |
case-05 | fail→pass | 15,763 | 17,408 | +10% | 1 | 1 | 0% | 2,174 | 2,877 | +32% | 0 | 0 | — |
case-06 | fail→fail | 9,368 | 12,289 | +31% | 1 | 1 | 0% | 1,286 | 1,878 | +46% | 0 | 0 | — |
case-07 | pass→pass | 21,625 | 18,755 | -13% | 1 | 1 | 0% | 3,147 | 3,099 | -2% | 0 | 0 | — |
case-08 | pass→pass | 40,823 | 31,440 | -23% | 1 | 1 | 0% | 5,753 | 4,843 | -16% | 0 | 0 | — |
case-09 | fail→pass | 39,689 | 30,807 | -22% | 1 | 1 | 0% | 6,197 | 5,002 | -19% | 0 | 0 | — |
case-10 | pass→pass | 25,483 | 21,045 | -17% | 1 | 1 | 0% | 3,777 | 3,295 | -13% | 0 | 0 | — |
case-11 | fail→pass | 17,199 | 16,433 | -4% | 1 | 1 | 0% | 2,546 | 2,684 | +5% | 0 | 0 | — |
case-12 | pass→pass | 19,102 | 35,097 | +84% | 1 | 1 | 0% | 2,642 | 5,289 | +100% | 0 | 0 | — |
case-13 | pass→pass | 25,478 | 24,452 | -4% | 1 | 1 | 0% | 3,618 | 3,690 | +2% | 0 | 0 | — |
case-14 | pass→pass | 22,937 | 32,059 | +40% | 1 | 1 | 0% | 3,377 | 5,108 | +51% | 0 | 0 | — |
case-15 | fail→pass | 17,985 | 22,812 | +27% | 1 | 1 | 0% | 2,525 | 3,424 | +36% | 0 | 0 | — |
case-16 | pass→pass | 31,035 | 29,183 | -6% | 1 | 1 | 0% | 4,595 | 4,779 | +4% | 0 | 0 | — |
case-17 | fail→fail | 15,925 | 15,872 | -0% | 1 | 1 | 0% | 2,264 | 2,637 | +16% | 0 | 0 | — |
case-18 | pass→pass | 22,680 | 33,505 | +48% | 1 | 1 | 0% | 3,348 | 5,403 | +61% | 0 | 0 | — |
case-19 | pass→pass | 24,649 | 40,072 | +63% | 1 | 1 | 0% | 3,422 | 6,150 | +80% | 0 | 0 | — |
case-20 | pass→fail | 15,294 | 24,583 | +61% | 1 | 1 | 0% | 2,507 | 4,112 | +64% | 0 | 0 | — |
case-21 | pass→fail | 15,328 | 14,381 | -6% | 1 | 1 | 0% | 2,561 | 2,622 | +2% | 0 | 0 | — |
case-22 | pass→fail | 11,534 | 13,311 | +15% | 1 | 1 | 0% | 2,139 | 2,745 | +28% | 0 | 0 | — |
case-23 | pass→fail | 13,784 | 23,292 | +69% | 1 | 1 | 0% | 1,977 | 3,660 | +85% | 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 +9 percentage points is the difference between those two pass rates over the 23 comparable cases. 4 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.