Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Legal document generation and compliance guidance for indie Apple developers. Covers privacy policies, terms of service, EULAs, GDPR/CCPA/DPDP compliance, and Apple App Store legal requirements. Use when user needs legal documents or compliance guidance.
.claude/skills/thomasmoreai-legal/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-07 | ✗→✓ | ▲ Improved | -63% | 0% |
| case-14 | ✗→✓ | ▲ Improved | 1% | 0% |
| case-04 | ✓→✓ | ＝ Same ✓ | -8% | 0% |
| case-05 | ✓→✓ | ＝ Same ✓ | -2% | 0% |
| case-06 | ✓→✓ | ＝ Same ✓ | -17% | 0% |
Legal document generation and compliance guidance for indie Apple developers. Generate privacy policies, terms of service, and EULAs tailored to your app's data practices and target markets.
Use this skill when the user:
Generate privacy policies, terms of service, and EULAs tailored to your app's data collection practices, third-party SDKs, and target regions. Includes Apple Privacy Nutrition Label mapping and hosting guidance.
privacy-policy/SKILL.md for the full generation workflowgenerators/consent-flow -- GDPR/CCPA consent UI and tracking consent managementgenerators/account-deletion -- Account deletion flow (App Store requirement)generators/permission-priming -- Pre-permission UI patterns (ATT priming)monetization/ -- Subscription terms and pricing disclosures| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-09 | fail→fail | 13,525 | 12,054 | -11% | 1 | 1 | 0% | 2,658 | 2,777 | +4% | 0 | 0 | — |
case-10 | fail→fail | 15,378 | 16,080 | +5% | 1 | 1 | 0% | 3,200 | 3,705 | +16% | 0 | 0 | — |
case-17 | fail→fail | 13,510 | 13,994 | +4% | 1 | 1 | 0% | 2,606 | 3,067 | +18% | 0 | 0 | — |
case-07 | fail→pass | 12,940 | 3,322 | -74% | 1 | 1 | 0% | 2,418 | 896 | -63% | 0 | 0 | — |
case-08 | fail→fail | 16,329 | 15,388 | -6% | 1 | 1 | 0% | 3,303 | 3,583 | +8% | 0 | 0 | — |
case-01 | fail→fail | 13,596 | 14,596 | +7% | 1 | 1 | 0% | 2,954 | 3,397 | +15% | 0 | 0 | — |
case-02 | fail→fail | 17,593 | 14,271 | -19% | 1 | 1 | 0% | 3,777 | 3,563 | -6% | 0 | 0 | — |
case-03 | fail→fail | 15,325 | 14,848 | -3% | 1 | 1 | 0% | 3,248 | 3,309 | +2% | 0 | 0 | — |
case-04 | pass→pass | 16,036 | 13,380 | -17% | 1 | 1 | 0% | 3,674 | 3,397 | -8% | 0 | 0 | — |
case-05 | pass→pass | 14,406 | 12,851 | -11% | 1 | 1 | 0% | 3,067 | 2,994 | -2% | 0 | 0 | — |
case-06 | pass→pass | 13,961 | 9,757 | -30% | 1 | 1 | 0% | 3,126 | 2,606 | -17% | 0 | 0 | — |
case-11 | fail→fail | 13,342 | 11,819 | -11% | 1 | 1 | 0% | 2,468 | 2,494 | +1% | 0 | 0 | — |
case-12 | fail→fail | 11,764 | 9,845 | -16% | 1 | 1 | 0% | 2,108 | 2,299 | +9% | 0 | 0 | — |
case-13 | pass→pass | 14,656 | 14,193 | -3% | 1 | 1 | 0% | 2,457 | 2,705 | +10% | 0 | 0 | — |
case-14 | fail→pass | 15,259 | 12,806 | -16% | 1 | 1 | 0% | 2,456 | 2,480 | +1% | 0 | 0 | — |
case-15 | fail→fail | 12,727 | 12,390 | -3% | 1 | 1 | 0% | 2,283 | 2,397 | +5% | 0 | 0 | — |
case-16 | fail→fail | 12,383 | 12,392 | +0% | 1 | 1 | 0% | 2,140 | 2,427 | +13% | 0 | 0 | — |
case-18 | pass→pass | 7,615 | 5,600 | -26% | 1 | 1 | 0% | 1,450 | 1,330 | -8% | 0 | 0 | — |
case-19 | fail→fail | 10,002 | 9,035 | -10% | 1 | 1 | 0% | 1,810 | 1,974 | +9% | 0 | 0 | — |
case-20 | pass→pass | 12,601 | 8,874 | -30% | 1 | 1 | 0% | 2,480 | 2,154 | -13% | 0 | 0 | — |
case-21 | fail→fail | 11,997 | 10,995 | -8% | 1 | 1 | 0% | 2,128 | 2,328 | +9% | 0 | 0 | — |
case-22 | fail→fail | 11,780 | 7,853 | -33% | 1 | 1 | 0% | 2,174 | 1,660 | -24% | 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 +9 percentage points is the difference between those two pass rates over the 22 comparable cases.
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.