Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Run a self-audit of a food establishment before the health inspector arrives, focused on the violations that actually close kitchens. Use when asked to prep for a health inspection, do a food-safety self-audit, avoid critical violations, or get ready for the health department. Produces a prioritized checklist organized by risk (critical/priority vs. non-critical), the temperature and hygiene fundamentals, a fix list with owners, and how to handle the inspector on the day.
.claude/skills/mohitagw15856-health-inspection-prep/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-12 | ✗→✓ | ▲ Improved | 24% | 0% |
| case-01 | ✗→✓ | ▲ Improved | 27% | 0% |
| case-03 | ✗→✓ | ▲ Improved | 25% | 0% |
| case-07 | ✗→✓ | ▲ Improved | 2% | 0% |
| case-08 | ✗→✓ | ▲ Improved | 2% | 0% |
Inspectors don't fail you on dust — they fail you on the handful of priority violations that make people sick: temperature abuse, cross-contamination, bare-hand contact, handwashing, and pest activity. This skill runs the self-audit the way an inspector scores it, so the critical stuff is fixed before they walk in.
Given the establishment type, produce the full self-audit — prioritize by public-health risk, not by what's easiest to see. Base it on standard food-code categories; note that local codes vary and the operator should confirm specifics.
Ask for (if not provided, else infer and label):
The violations that shut kitchens — each with the standard and a pass/fail check:
Facility, labeling, storage, cleaning cadence.
| Item | Risk | Fix | Owner | By when | |---|---|---|---|---|
How to receive the inspector: cooperate, don't argue, know where the logs and permits are, correct-on-the-spot where possible, and how to respond to a violation professionally.
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-12 | fail→pass | 18,952 | 19,854 | +5% | 1 | 1 | 0% | 3,013 | 3,725 | +24% | 0 | 0 | — |
case-01 | fail→pass | 20,585 | 20,366 | -1% | 1 | 1 | 0% | 3,529 | 4,477 | +27% | 0 | 0 | — |
case-02 | fail→fail | 17,233 | 23,143 | +34% | 1 | 1 | 0% | 3,903 | 4,561 | +17% | 0 | 0 | — |
case-03 | fail→pass | 18,334 | 24,812 | +35% | 1 | 1 | 0% | 3,926 | 4,893 | +25% | 0 | 0 | — |
case-04 | pass→pass | 17,959 | 20,951 | +17% | 1 | 1 | 0% | 3,169 | 4,309 | +36% | 0 | 0 | — |
case-05 | pass→fail | 23,700 | 28,080 | +18% | 1 | 1 | 0% | 4,068 | 5,449 | +34% | 0 | 0 | — |
case-06 | pass→pass | 14,753 | 22,925 | +55% | 1 | 1 | 0% | 3,797 | 5,441 | +43% | 0 | 0 | — |
case-07 | fail→pass | 20,865 | 17,068 | -18% | 1 | 1 | 0% | 3,435 | 3,490 | +2% | 0 | 0 | — |
case-08 | fail→pass | 21,720 | 19,982 | -8% | 1 | 1 | 0% | 3,958 | 4,035 | +2% | 0 | 0 | — |
case-09 | pass→pass | 16,803 | 15,816 | -6% | 1 | 1 | 0% | 2,990 | 3,721 | +24% | 0 | 0 | — |
case-10 | fail→pass | 18,529 | 17,038 | -8% | 1 | 1 | 0% | 3,094 | 3,740 | +21% | 0 | 0 | — |
case-11 | pass→pass | 25,042 | 20,640 | -18% | 1 | 1 | 0% | 4,069 | 4,021 | -1% | 0 | 0 | — |
case-13 | fail→fail | 15,111 | 17,181 | +14% | 1 | 1 | 0% | 2,640 | 3,391 | +28% | 0 | 0 | — |
case-14 | fail→fail | 20,952 | 20,107 | -4% | 1 | 1 | 0% | 3,295 | 3,785 | +15% | 0 | 0 | — |
case-15 | pass→pass | 17,049 | 17,453 | +2% | 1 | 1 | 0% | 3,051 | 3,426 | +12% | 0 | 0 | — |
case-16 | pass→pass | 21,660 | 20,195 | -7% | 1 | 1 | 0% | 3,362 | 3,825 | +14% | 0 | 0 | — |
case-17 | fail→pass | 16,778 | 18,725 | +12% | 1 | 1 | 0% | 2,632 | 3,663 | +39% | 0 | 0 | — |
case-18 | fail→pass | 21,300 | 18,946 | -11% | 1 | 1 | 0% | 3,626 | 3,687 | +2% | 0 | 0 | — |
case-19 | pass→pass | 18,756 | 21,196 | +13% | 1 | 1 | 0% | 3,342 | 4,386 | +31% | 0 | 0 | — |
case-20 | fail→pass | 21,356 | 20,605 | -4% | 1 | 1 | 0% | 3,705 | 4,063 | +10% | 0 | 0 | — |
case-21 | pass→pass | 20,543 | 19,859 | -3% | 1 | 1 | 0% | 3,663 | 4,215 | +15% | 0 | 0 | — |
case-22 | fail→pass | 20,757 | 17,848 | -14% | 1 | 1 | 0% | 3,266 | 3,500 | +7% | 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 +41 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.