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 until reaching shared understanding, resolving each branch of the decision tree. Use when user wants to stress-test a plan, get grilled on their design, or mentions "grill me".
.claude/skills/alirezarezvani-grill-me/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-03 | ✗→✓ | ▲ Improved | 33% | 0% |
| case-06 | ✗→✓ | ▲ Improved | 94% | 0% |
| case-07 | ✗→✓ | ▲ Improved | 68% | 0% |
| case-14 | ✗→✓ | ▲ Improved | -15% | 0% |
| case-15 | ✗→✓ | ▲ Improved | -44% | 0% |
> Derived from Matt Pocock's grill-me (MIT). Matt's interview discipline preserved verbatim. Additions: extraction + question + session tools + references + cs- wrapper (see references/companion_tooling.md).
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.
If a question can be answered by exploring the codebase, explore the codebase instead.
grep / Read resolves it, do that first. Saves a turn.scripts/decision_tree_extractor.py to extract branches.scripts/question_generator.py to produce the question list with recommendations.scripts/grill_session_tracker.py --action start.Per question turn:
Q[i]/[total]: [question]
Recommended answer: [your call + 1-sentence rationale]
(Or: I explored the codebase and found [evidence]. Confirm?)See references/companion_tooling.md. Tools: extractor + generator + tracker. Agent: cs-grill-master. Command: /cs:grill-me.
Version: 1.0.0 Derived: Matt Pocock (MIT) + this repo's wrapper
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→fail | 4,313 | 3,986 | -8% | 1 | 1 | 0% | 302 | 819 | +171% | 0 | 0 | — |
case-02 | fail→fail | 7,154 | 5,118 | -28% | 1 | 1 | 0% | 1,233 | 845 | -31% | 0 | 0 | — |
case-03 | fail→pass | 6,644 | 5,326 | -20% | 1 | 1 | 0% | 1,154 | 1,535 | +33% | 0 | 0 | — |
case-04 | pass→fail | 7,798 | 2,714 | -65% | 1 | 1 | 0% | 1,449 | 983 | -32% | 0 | 0 | — |
case-05 | pass→pass | 8,490 | 3,175 | -63% | 1 | 1 | 0% | 1,569 | 847 | -46% | 0 | 0 | — |
case-06 | fail→pass | 3,318 | 2,671 | -19% | 1 | 1 | 0% | 554 | 1,073 | +94% | 0 | 0 | — |
case-07 | fail→pass | 6,079 | 7,396 | +22% | 1 | 1 | 0% | 1,139 | 1,908 | +68% | 0 | 0 | — |
case-08 | pass→pass | 2,704 | 1,890 | -30% | 1 | 1 | 0% | 483 | 892 | +85% | 0 | 0 | — |
case-09 | pass→pass | 5,437 | 2,699 | -50% | 1 | 1 | 0% | 948 | 1,043 | +10% | 0 | 0 | — |
case-10 | pass→pass | 1,940 | 2,794 | +44% | 1 | 1 | 0% | 304 | 968 | +218% | 0 | 0 | — |
case-11 | fail→fail | 6,412 | 2,698 | -58% | 1 | 1 | 0% | 1,179 | 1,045 | -11% | 0 | 0 | — |
case-12 | pass→pass | 4,771 | 4,126 | -14% | 1 | 1 | 0% | 830 | 1,401 | +69% | 0 | 0 | — |
case-13 | fail→fail | 4,661 | 1,967 | -58% | 1 | 1 | 0% | 750 | 859 | +15% | 0 | 0 | — |
case-14 | fail→pass | 6,665 | 2,656 | -60% | 1 | 1 | 0% | 1,171 | 994 | -15% | 0 | 0 | — |
case-15 | fail→pass | 8,015 | 1,464 | -82% | 1 | 1 | 0% | 1,335 | 754 | -44% | 0 | 0 | — |
case-16 | fail→pass | 4,614 | 2,968 | -36% | 1 | 1 | 0% | 838 | 1,135 | +35% | 0 | 0 | — |
case-17 | pass→pass | 5,858 | 3,252 | -44% | 1 | 1 | 0% | 1,035 | 1,152 | +11% | 0 | 0 | — |
case-18 | fail→pass | 1,479 | 2,818 | +91% | 1 | 1 | 0% | 230 | 1,087 | +373% | 0 | 0 | — |
case-19 | pass→pass | 6,687 | 1,636 | -76% | 1 | 1 | 0% | 1,154 | 800 | -31% | 0 | 0 | — |
case-20 | pass→fail | 12,183 | 8,179 | -33% | 1 | 1 | 0% | 2,091 | 1,937 | -7% | 0 | 0 | — |
case-21 | pass→pass | 8,787 | 8,825 | +0% | 1 | 1 | 0% | 1,624 | 2,019 | +24% | 0 | 0 | — |
case-22 | pass→fail | 7,974 | 4,527 | -43% | 1 | 1 | 0% | 1,940 | 1,443 | -26% | 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, and 20 counted toward the lift figure. The other 2 produced results that are not comparable between the two arms, so they are excluded from the headline rather than averaged into it. The headline lift of +18 percentage points is the difference between those two pass rates over the 20 comparable cases. 3 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.