Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Write a factory acceptance test (FAT) plan or report — test coverage matrix against spec, AQL sampling plan, pass/fail criteria, golden-sample handling, deviation log, and sign-off structure. Use when asked to write a FAT plan, define outgoing quality inspection, set AQL levels, prepare for a factory acceptance or pre-shipment inspection, or document FAT results. Produces a complete FAT plan or report with sampling tables, defect classification, and a sign-off block.
.claude/skills/mohitagw15856-factory-acceptance-test/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-01 | ✗→✓ | ▲ Improved | -7% | 0% |
| case-02 | ✗→✓ | ▲ Improved | 73% | 0% |
| case-03 | ✗→✓ | ▲ Improved | -1% | 0% |
| case-07 | ✗→✓ | ▲ Improved | 34% | 0% |
| case-08 | ✗→✓ | ▲ Improved | 29% | 0% |
A FAT is the last moment a defect is the factory's problem instead of yours. This skill writes the plan (or the report) with the parts that actually get argued over at the line: which spec clause each test covers, how many units get pulled and at what AQL, what exactly fails a lot, and who signs — so acceptance is a decision with a name on it, not a vibe at the end of a factory visit.
Ask for these if not provided; if the spec is thin, draft the matrix from the product description and mark rows [spec clause to confirm]:
Defect classes and default AQLs (ANSI/ASQ Z1.4, General Inspection Level II, normal inspection — the consumer-electronics defaults; adjust with reason):
| Class | Definition | Default AQL | |---|---|---| | Critical | Safety hazard, regulatory violation, data loss | 0 — any occurrence rejects the lot | | Major | Product fails to function, or defect the user will certainly notice and return | 1.0 (tighten to 0.65 for premium/first lots) | | Minor | Cosmetic or workmanship issue within limit samples' tolerance but noted | 2.5 (relax to 4.0 for bulk/industrial) |
From lot size + Level II, derive the sample-size code letter and accept/reject numbers from the Z1.4 tables; state them explicitly in the plan (e.g. "Lot 3,000 → code K → n=125; Major Ac=3/Re=4"). Switch to tightened inspection after 2 of 5 consecutive lots rejected; reduced only with sustained history.
Golden samples. Two signed, serialised golden samples minimum — one held at the factory line, one at the buyer. Sealed, dated, with an expiry/refresh rule (refresh on any ECO that changes fit/finish/function). Cosmetic judgement is against the limit-sample boundary set, never against memory.
Deviations. Any test performed differently than planned, any borderline judgement, and any use-as-is decision goes in the deviation log — numbered, with disposition and approver.
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→pass | 29,642 | 22,148 | -25% | 1 | 1 | 0% | 5,827 | 5,431 | -7% | 0 | 0 | — |
case-02 | fail→pass | 15,044 | 20,736 | +38% | 1 | 1 | 0% | 2,819 | 4,863 | +73% | 0 | 0 | — |
case-03 | fail→pass | 29,613 | 22,847 | -23% | 1 | 1 | 0% | 5,224 | 5,179 | -1% | 0 | 0 | — |
case-04 | pass→pass | 16,085 | 23,502 | +46% | 1 | 1 | 0% | 2,756 | 4,895 | +78% | 0 | 0 | — |
case-05 | pass→fail | 20,925 | 21,585 | +3% | 1 | 1 | 0% | 3,478 | 4,417 | +27% | 0 | 0 | — |
case-06 | pass→pass | 15,894 | 24,094 | +52% | 1 | 1 | 0% | 2,674 | 5,069 | +90% | 0 | 0 | — |
case-07 | fail→pass | 14,597 | 12,850 | -12% | 1 | 1 | 0% | 2,468 | 3,300 | +34% | 0 | 0 | — |
case-08 | fail→pass | 12,306 | 10,260 | -17% | 1 | 1 | 0% | 2,355 | 3,040 | +29% | 0 | 0 | — |
case-09 | pass→pass | 10,447 | 12,552 | +20% | 1 | 1 | 0% | 1,853 | 3,362 | +81% | 0 | 0 | — |
case-10 | fail→pass | 18,920 | 16,218 | -14% | 1 | 1 | 0% | 2,919 | 3,718 | +27% | 0 | 0 | — |
case-11 | pass→pass | 14,474 | 15,143 | +5% | 1 | 1 | 0% | 2,497 | 3,659 | +47% | 0 | 0 | — |
case-12 | pass→pass | 13,538 | 10,774 | -20% | 1 | 1 | 0% | 1,971 | 2,785 | +41% | 0 | 0 | — |
case-13 | fail→pass | 13,485 | 17,790 | +32% | 1 | 1 | 0% | 2,265 | 3,990 | +76% | 0 | 0 | — |
case-14 | pass→pass | 32,013 | 10,873 | -66% | 1 | 1 | 0% | 1,870 | 3,018 | +61% | 0 | 0 | — |
case-15 | fail→pass | 11,213 | 17,339 | +55% | 1 | 1 | 0% | 1,879 | 3,803 | +102% | 0 | 0 | — |
case-16 | fail→pass | 43,758 | 10,863 | -75% | 1 | 1 | 0% | 899 | 2,910 | +224% | 0 | 0 | — |
case-17 | fail→pass | 22,421 | 12,989 | -42% | 1 | 1 | 0% | 2,893 | 3,566 | +23% | 0 | 0 | — |
case-18 | pass→pass | 12,469 | 22,256 | +78% | 1 | 1 | 0% | 2,289 | 4,810 | +110% | 0 | 0 | — |
case-19 | fail→pass | 9,026 | 8,245 | -9% | 1 | 1 | 0% | 1,494 | 2,402 | +61% | 0 | 0 | — |
case-20 | fail→pass | 7,469 | 3,509 | -53% | 1 | 1 | 0% | 1,346 | 1,768 | +31% | 0 | 0 | — |
case-21 | fail→fail | 7,900 | 3,863 | -51% | 1 | 1 | 0% | 1,434 | 1,765 | +23% | 0 | 0 | — |
case-22 | fail→pass | 9,895 | 14,139 | +43% | 1 | 1 | 0% | 1,715 | 3,434 | +100% | 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 21 counted toward the lift figure. The other 1 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 +55 percentage points is the difference between those two pass rates over the 21 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.