Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Runs a stateful grilling session that produces workflow specs for workflows the user wants to build. Use when the user says "loop me" or wants workflow specs drawn out of them.
.claude/skills/fradser-loop-me/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-02 | ✗→✓ | ▲ Improved | -4% | 0% |
| case-03 | ✗→✓ | ▲ Improved | -6% | 0% |
| case-07 | ✗→✓ | ▲ Improved | -35% | 0% |
| case-12 | ✗→✓ | ▲ Improved | 36% | 0% |
| case-16 | ✗→✓ | ▲ Improved | 29% | 0% |
Run a stateful /mattpocock:grilling session whose only output is workflow specs. Use the grilling discipline — relentless, one question at a time, a recommended answer attached to each — aimed at the vocabulary and goal below. Create, edit, and delete specs as the grilling resolves things.
A loop is a recurring pattern in the user's life: their career, their week, their morning, a single repeated activity. Picturing a life as loops within loops reveals how predictable its activities really are — which is what makes them worth delegating. Use the lens to find loops worth specifying, and propose ones the user hasn't noticed.
A workflow is the spec of one loop, made real. You run a workflow on a loop — the loop is its running instantiation. Workflows live in workflows/*.md and are the source of truth.
A shared language, reached for only when a workflow calls for it — never a checklist. Mandate nothing structural: a workflow needs no AI, no checkpoint, and no schedule unless the grilling shows it does.
A workflow spec is done when an implementer agent could build it without asking a single question. Grill until then; nothing is done while a question remains.
workflows/*.md — one spec per workflow.NOTES.md — raw notes on the user's world: the tools they use, the channels they process, and their own terminology for both. When it is empty or thin, interview them about their world before specifying anything. Sharpen fuzzy terms into canonical ones as they surface, and record them here.| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→fail | 11,448 | 27,196 | +138% | 1 | 1 | 0% | 1,790 | 869 | -51% | 0 | 0 | — |
case-02 | fail→pass | 8,567 | 5,685 | -34% | 1 | 1 | 0% | 1,374 | 1,321 | -4% | 0 | 0 | — |
case-03 | fail→pass | 9,195 | 5,510 | -40% | 1 | 1 | 0% | 1,474 | 1,381 | -6% | 0 | 0 | — |
case-04 | pass→fail | 15,937 | 20,422 | +28% | 1 | 1 | 0% | 3,359 | 3,793 | +13% | 0 | 0 | — |
case-05 | pass→fail | 20,246 | 4,374 | -78% | 1 | 1 | 0% | 4,123 | 865 | -79% | 0 | 0 | — |
case-19 | pass→fail | 11,600 | 3,961 | -66% | 1 | 1 | 0% | 1,926 | 740 | -62% | 0 | 0 | — |
case-06 | pass→fail | 14,158 | 16,377 | +16% | 1 | 1 | 0% | 2,118 | 3,057 | +44% | 0 | 0 | — |
case-07 | fail→pass | 17,829 | 7,898 | -56% | 1 | 1 | 0% | 2,818 | 1,827 | -35% | 0 | 0 | — |
case-08 | pass→fail | 16,548 | 5,152 | -69% | 1 | 1 | 0% | 2,452 | 829 | -66% | 0 | 0 | — |
case-09 | pass→pass | 11,112 | 10,179 | -8% | 1 | 1 | 0% | 1,717 | 2,242 | +31% | 0 | 0 | — |
case-10 | pass→pass | 12,702 | 11,733 | -8% | 1 | 1 | 0% | 2,069 | 2,041 | -1% | 0 | 0 | — |
case-11 | pass→pass | 9,412 | 6,405 | -32% | 1 | 1 | 0% | 1,778 | 1,642 | -8% | 0 | 0 | — |
case-12 | fail→pass | 11,369 | 10,378 | -9% | 1 | 1 | 0% | 1,770 | 2,415 | +36% | 0 | 0 | — |
case-13 | pass→pass | 13,474 | 8,111 | -40% | 1 | 1 | 0% | 2,002 | 1,596 | -20% | 0 | 0 | — |
case-14 | pass→pass | 7,352 | 6,386 | -13% | 1 | 1 | 0% | 1,297 | 1,627 | +25% | 0 | 0 | — |
case-15 | fail→fail | 14,728 | 5,250 | -64% | 1 | 1 | 0% | 2,333 | 1,321 | -43% | 0 | 0 | — |
case-16 | fail→pass | 10,844 | 10,447 | -4% | 1 | 1 | 0% | 1,851 | 2,388 | +29% | 0 | 0 | — |
case-17 | pass→fail | 11,583 | 3,379 | -71% | 1 | 1 | 0% | 2,001 | 781 | -61% | 0 | 0 | — |
case-18 | fail→pass | 19,575 | 3,782 | -81% | 1 | 1 | 0% | 1,821 | 1,141 | -37% | 0 | 0 | — |
case-20 | pass→pass | 11,995 | 7,291 | -39% | 1 | 1 | 0% | 1,985 | 1,699 | -14% | 0 | 0 | — |
case-21 | pass→pass | 9,604 | 10,611 | +10% | 1 | 1 | 0% | 1,589 | 1,599 | +1% | 0 | 0 | — |
case-22 | pass→pass | 8,291 | 5,388 | -35% | 1 | 1 | 0% | 1,289 | 1,356 | +5% | 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 17 counted toward the lift figure. The other 5 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 0 percentage points is the difference between those two pass rates over the 17 comparable cases. 6 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.