Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Interview the user relentlessly about a plan or design. Use when the user wants to stress-test a plan before building, or uses any 'grill' trigger phrases.
.claude/skills/sickn33-grilling/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-18 | ✗→✓ | ▲ Improved | -73% | 0% |
| case-05 | ✗→✓ | ▲ Improved | -64% | 0% |
| case-01 | ✗→✓ | ▲ Improved | -33% | 0% |
| case-02 | ✗→✓ | ▲ Improved | -52% | 0% |
| case-03 | ✗→✓ | ▲ Improved | -35% | 0% |
Use when this workflow matches the user request: Interview the user relentlessly about a plan or design. Use when the user wants to stress-test a plan before building, or uses any 'grill' trigger phrases.
_Source: mattpocock/skills (MIT)._Interview me relentlessly about every aspect of this plan until we reach a shared understanding. Walk down each branch of the design tree, resolving dependencies between decisions one-by-one. For each question, provide your recommended answer.
Ask the questions one at a time, waiting for feedback on each question before continuing. Asking multiple questions at once is bewildering.
If a question can be answered by exploring the codebase, explore the codebase instead.
User request:
> Use @grilling for this task: Interview the user relentlessly about a plan or design.
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-18 | fail→pass | 24,156 | 5,833 | -76% | 1 | 1 | 0% | 3,731 | 1,004 | -73% | 0 | 0 | — |
case-05 | fail→pass | 16,603 | 4,886 | -71% | 1 | 1 | 0% | 2,355 | 850 | -64% | 0 | 0 | — |
case-01 | fail→pass | 8,624 | 4,370 | -49% | 1 | 1 | 0% | 1,214 | 810 | -33% | 0 | 0 | — |
case-02 | fail→pass | 19,477 | 6,132 | -69% | 1 | 1 | 0% | 2,330 | 1,117 | -52% | 0 | 0 | — |
case-03 | fail→pass | 9,055 | 4,402 | -51% | 1 | 1 | 0% | 1,236 | 809 | -35% | 0 | 0 | — |
case-04 | fail→fail | 13,130 | 5,599 | -57% | 1 | 1 | 0% | 1,923 | 714 | -63% | 0 | 0 | — |
case-06 | fail→pass | 15,777 | 5,715 | -64% | 1 | 1 | 0% | 2,205 | 945 | -57% | 0 | 0 | — |
case-07 | fail→pass | 38,923 | 5,693 | -85% | 1 | 1 | 0% | 2,728 | 979 | -64% | 0 | 0 | — |
case-08 | fail→pass | 9,416 | 6,000 | -36% | 1 | 1 | 0% | 1,306 | 1,141 | -13% | 0 | 0 | — |
case-09 | fail→pass | 17,169 | 5,467 | -68% | 1 | 1 | 0% | 2,443 | 912 | -63% | 0 | 0 | — |
case-10 | fail→pass | 18,059 | 6,304 | -65% | 1 | 1 | 0% | 2,599 | 1,132 | -56% | 0 | 0 | — |
case-11 | fail→pass | 16,259 | 4,967 | -69% | 1 | 1 | 0% | 2,349 | 900 | -62% | 0 | 0 | — |
case-12 | fail→pass | 16,334 | 4,378 | -73% | 1 | 1 | 0% | 2,399 | 819 | -66% | 0 | 0 | — |
case-13 | fail→pass | 22,459 | 5,498 | -76% | 1 | 1 | 0% | 3,362 | 1,046 | -69% | 0 | 0 | — |
case-14 | fail→pass | 15,482 | 7,449 | -52% | 1 | 1 | 0% | 2,217 | 940 | -58% | 0 | 0 | — |
case-15 | fail→pass | 17,002 | 6,356 | -63% | 1 | 1 | 0% | 2,416 | 1,084 | -55% | 0 | 0 | — |
case-16 | fail→fail | 24,201 | 5,029 | -79% | 1 | 1 | 0% | 3,562 | 930 | -74% | 0 | 0 | — |
case-17 | fail→fail | 19,878 | 4,734 | -76% | 1 | 1 | 0% | 2,847 | 887 | -69% | 0 | 0 | — |
case-19 | fail→fail | 16,961 | 4,581 | -73% | 1 | 1 | 0% | 2,474 | 879 | -64% | 0 | 0 | — |
case-20 | pass→pass | 9,388 | 8,139 | -13% | 1 | 1 | 0% | 1,673 | 1,610 | -4% | 0 | 0 | — |
case-21 | pass→pass | 19,427 | 14,160 | -27% | 1 | 1 | 0% | 2,929 | 2,278 | -22% | 0 | 0 | — |
case-22 | pass→pass | 11,495 | 11,825 | +3% | 1 | 1 | 0% | 2,216 | 2,500 | +13% | 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 +68 percentage points is the difference between those two pass rates over the 22 comparable cases.
The publisher has shipped newer versions since this run, so these numbers describe v1, not the version currently listed.
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.