Install any skill in seconds. Free to start, no credit card required.
Get Started Free →This skill should be used when implementing features with parallel exploration or competition. Triggers on "build", "create", "implement", "try both approaches", "compare implementations". Routes to omakase-off (entry gate for design exploration) or cookoff (exit gate for parallel implementation).
.claude/skills/aiskillstore-test-kitchen/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-09 | ✗→✓ | ▲ Improved | -8% | 0% |
| case-15 | ✗→✓ | ▲ Improved | 31% | 0% |
| case-02 | ✗→✓ | ▲ Improved | -30% | 0% |
| case-03 | ✗→✓ | ▲ Improved | -40% | 0% |
| case-04 | ✗→✓ | ▲ Improved | -9% | 0% |
Parallel implementation framework with two gate skills:
| Skill | Gate | Trigger | |-------|------|---------| | test-kitchen:omakase-off | Entry | FIRST on any build/create/implement request | | test-kitchen:cookoff | Exit | At design→implementation transition |
"Build X" / "Create Y" / "Implement Z"
↓
┌─────────────────────────────────────┐
│ OMAKASE-OFF (entry gate) │
│ Wraps brainstorming │
│ │
│ Choice: │
│ 1. Brainstorm together │
│ 2. Omakase (3-5 parallel designs) │
└─────────────────────────────────────┘
↓
[Brainstorming / Design phase]
↓
Design complete, "let's implement"
↓
┌─────────────────────────────────────┐
│ COOKOFF (exit gate) │
│ Wraps implementation │
│ │
│ Choice: │
│ 1. Cookoff (2-5 parallel agents) │
│ 2. Single subagent │
│ 3. Local implementation │
└─────────────────────────────────────┘
↓
[Implementation]Skills need aggressive triggers to work. They can't passively detect "uncertainty" or "readiness" - they must claim specific moments in the conversation flow.
Trigger 1: BEFORE brainstorming
Trigger 2: DURING brainstorming (slot detection)
Trigger 3: Explicitly requested
If user picks "Omakase" at the entry gate:
Best for: "I'm flexible, show me options in working code"
If user picks "Cookoff" at the exit gate:
Best for: "I want to see different implementation approaches"
| | Omakase-off | Cookoff | |-|-------------|---------| | Gate | Entry (before/during brainstorming) | Exit (after design) | | Question | HOW to explore? | HOW to implement? | | Parallel on | Different DESIGNS | Same design, different PLANS | | Triggers | Build request, indecision detection, explicit | "let's implement" signal | | Skips | Brainstorming (optional via short-circuit) | Nothing - always after design |
When omakase-off delegates to brainstorming, it passively tracks architectural decisions where user shows uncertainty:
Detection signals:
Slot classification: | Type | Examples | Worth exploring? | |------|----------|------------------| | Architectural | Storage engine, framework, auth method | Yes - different code paths | | Trivial | File location, naming, config format | No - easy to change |
At end of brainstorming:
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-09 | fail→pass | 14,410 | 6,155 | -57% | 1 | 1 | 0% | 2,131 | 1,957 | -8% | 0 | 0 | — |
case-15 | fail→pass | 15,687 | 16,685 | +6% | 1 | 1 | 0% | 2,460 | 3,219 | +31% | 0 | 0 | — |
case-14 | pass→pass | 13,168 | 5,802 | -56% | 1 | 1 | 0% | 1,663 | 2,040 | +23% | 0 | 0 | — |
case-01 | pass→pass | 12,806 | 5,063 | -60% | 1 | 1 | 0% | 1,886 | 1,782 | -6% | 0 | 0 | — |
case-02 | fail→pass | 17,179 | 4,650 | -73% | 1 | 1 | 0% | 2,565 | 1,804 | -30% | 0 | 0 | — |
case-03 | fail→pass | 15,316 | 2,683 | -82% | 1 | 1 | 0% | 2,370 | 1,412 | -40% | 0 | 0 | — |
case-04 | fail→pass | 11,307 | 3,951 | -65% | 1 | 1 | 0% | 1,727 | 1,565 | -9% | 0 | 0 | — |
case-05 | fail→pass | 16,917 | 4,854 | -71% | 1 | 1 | 0% | 2,880 | 1,656 | -43% | 0 | 0 | — |
case-06 | fail→pass | 13,839 | 3,943 | -72% | 1 | 1 | 0% | 2,158 | 1,659 | -23% | 0 | 0 | — |
case-07 | fail→pass | 16,106 | 4,706 | -71% | 1 | 1 | 0% | 2,453 | 1,634 | -33% | 0 | 0 | — |
case-08 | fail→pass | 16,447 | 3,754 | -77% | 1 | 1 | 0% | 2,543 | 1,602 | -37% | 0 | 0 | — |
case-10 | pass→pass | 7,251 | 3,659 | -50% | 1 | 1 | 0% | 1,132 | 1,556 | +37% | 0 | 0 | — |
case-11 | pass→pass | 10,367 | 4,312 | -58% | 1 | 1 | 0% | 1,635 | 1,637 | +0% | 0 | 0 | — |
case-12 | pass→pass | 18,407 | 6,292 | -66% | 1 | 1 | 0% | 3,445 | 1,980 | -43% | 0 | 0 | — |
case-13 | fail→fail | 7,227 | 2,085 | -71% | 1 | 1 | 0% | 1,023 | 1,340 | +31% | 0 | 0 | — |
case-16 | fail→pass | 10,533 | 2,263 | -79% | 1 | 1 | 0% | 1,498 | 1,337 | -11% | 0 | 0 | — |
case-17 | pass→pass | 8,868 | 2,104 | -76% | 1 | 1 | 0% | 1,220 | 1,319 | +8% | 0 | 0 | — |
case-18 | pass→pass | 8,243 | 3,678 | -55% | 1 | 1 | 0% | 1,192 | 1,601 | +34% | 0 | 0 | — |
case-19 | pass→pass | 9,845 | 3,160 | -68% | 1 | 1 | 0% | 1,471 | 1,447 | -2% | 0 | 0 | — |
case-20 | fail→fail | 10,818 | 11,965 | +11% | 1 | 1 | 0% | 1,213 | 2,120 | +75% | 0 | 0 | — |
case-21 | fail→pass | 4,618 | 8,057 | +74% | 1 | 1 | 0% | 766 | 2,267 | +196% | 0 | 0 | — |
case-22 | pass→fail | 15,222 | 6,141 | -60% | 1 | 1 | 0% | 3,304 | 1,919 | -42% | 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 +45 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.