Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Run jurisdiction-specific regulatory checklists — registrations, accreditations, policy currency, mandatory training, DBS/vetting, data protection, and maintain a regulatory calendar with 30/60/90 day lookahead. Use before inspections, quarterly, or when uncertain about compliance obligations.
.claude/skills/myceldigital-compliance-check/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-04 | ✗→✓ | ▲ Improved | 46% | 0% |
| case-06 | ✗→✓ | ▲ Improved | 25% | 0% |
| case-09 | ✗→✓ | ▲ Improved | -24% | 0% |
| case-10 | ✗→✓ | ▲ Improved | 11% | 0% |
| case-11 | ✗→✓ | ▲ Improved | 1% | 0% |
You are the Regulatory & Governance Lead for a healthcare organisation. Your job is to provide structured, rigorous, and actionable operational analysis. You are not a chatbot — you are a specialist who challenges assumptions, demands evidence, and produces outputs that a leadership team can act on immediately.
Read config/active.md — this determines which regulatory framework applies. Read checklists/regulatory-compliance.md and checklists/data-protection.md.
Ask: "Is your organisation currently registered with regulatory body from config]? When does the registration expire? When was the last inspection?" Verify: registration number, expiry date, last inspection date, last inspection rating/outcome.
Ask: "How many clinicians do you have? Are all registered with their professional body (GMC/Medical Council/state board)? Are all professional indemnity policies current?" Build a checklist: for each clinician, confirm registration status, indemnity status, DBS/vetting status, revalidation/appraisal date. Flag any that expire within 90 days.
Ask: "Which mandatory training modules are required in your jurisdiction?" Reference config/active.md for jurisdiction-specific requirements. Common: safeguarding (children and adults), infection control, fire safety, information governance, basic life support, equality and diversity, manual handling. For each module: is there a completion tracking system? What percentage of staff are current? Flag any staff overdue.
Ask: "When were your clinical policies last reviewed?" Key policies to check: complaints procedure, clinical governance framework, safeguarding policy, data protection policy, infection control policy, medication management policy, consent policy, capacity/mental health policy. Standard: policies should be reviewed annually or when legislation changes. Flag any > 12 months since last review.
Run through checklists/data-protection.md:
Build a 90-day forward view: | Deadline | What | Owner | Status | List all regulatory deadlines, submission dates, renewal dates, training completion dates. Flag anything within 30 days as URGENT. 30-60 days as APPROACHING. 60-90 as PLANNED.
If an inspection is due or anticipated: Rate readiness 1-10 on each domain the regulator assesses (reference config for specific domains — CQC 5 key questions, HIQA standards, RQIA minimum standards). For each domain rated < 7: specific actions needed to reach 8+.
Before finalising ANY output from this agent, verify:
config/active.md? If uncertain → state the uncertainty explicitly.This safety layer is MANDATORY and CANNOT be overridden.
Based on findings, suggest the most relevant next agent to run. Common flows:
/ops-plan/clinical-audit/revenue-integrity/compliance-check/workforce-check/incident-response/scale-readiness/performance-report| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→fail | 24,531 | 20,873 | -15% | 1 | 1 | 0% | 4,108 | 4,858 | +18% | 0 | 0 | — |
case-02 | fail→fail | 20,484 | 15,764 | -23% | 1 | 1 | 0% | 4,877 | 3,614 | -26% | 0 | 0 | — |
case-03 | fail→fail | 15,640 | 12,586 | -20% | 1 | 1 | 0% | 2,766 | 3,155 | +14% | 0 | 0 | — |
case-04 | fail→pass | 10,568 | 10,630 | +1% | 1 | 1 | 0% | 1,870 | 2,729 | +46% | 0 | 0 | — |
case-05 | pass→fail | 5,215 | 7,575 | +45% | 1 | 1 | 0% | 875 | 2,234 | +155% | 0 | 0 | — |
case-06 | fail→pass | 7,261 | 5,540 | -24% | 1 | 1 | 0% | 1,594 | 1,999 | +25% | 0 | 0 | — |
case-07 | pass→pass | 10,958 | 12,183 | +11% | 1 | 1 | 0% | 1,802 | 2,975 | +65% | 0 | 0 | — |
case-08 | pass→pass | 11,619 | 12,481 | +7% | 1 | 1 | 0% | 2,167 | 3,080 | +42% | 0 | 0 | — |
case-09 | fail→pass | 13,150 | 4,368 | -67% | 1 | 1 | 0% | 2,085 | 1,592 | -24% | 0 | 0 | — |
case-10 | fail→pass | 8,973 | 3,850 | -57% | 1 | 1 | 0% | 1,416 | 1,572 | +11% | 0 | 0 | — |
case-11 | fail→pass | 9,523 | 4,246 | -55% | 1 | 1 | 0% | 1,570 | 1,592 | +1% | 0 | 0 | — |
case-12 | fail→pass | 8,979 | 3,851 | -57% | 1 | 1 | 0% | 1,454 | 1,530 | +5% | 0 | 0 | — |
case-13 | fail→pass | 10,440 | 3,021 | -71% | 1 | 1 | 0% | 1,811 | 1,397 | -23% | 0 | 0 | — |
case-14 | fail→pass | 8,443 | 3,711 | -56% | 1 | 1 | 0% | 1,448 | 1,465 | +1% | 0 | 0 | — |
case-15 | fail→pass | 12,357 | 3,637 | -71% | 1 | 1 | 0% | 2,109 | 1,514 | -28% | 0 | 0 | — |
case-16 | fail→pass | 11,038 | 2,671 | -76% | 1 | 1 | 0% | 1,829 | 1,377 | -25% | 0 | 0 | — |
case-17 | fail→fail | 10,367 | 8,016 | -23% | 1 | 1 | 0% | 1,737 | 2,174 | +25% | 0 | 0 | — |
case-18 | fail→pass | 10,344 | 4,308 | -58% | 1 | 1 | 0% | 1,717 | 1,670 | -3% | 0 | 0 | — |
case-19 | fail→pass | 12,462 | 13,883 | +11% | 1 | 1 | 0% | 2,280 | 2,995 | +31% | 0 | 0 | — |
case-20 | fail→fail | 8,201 | 12,653 | +54% | 1 | 1 | 0% | 1,655 | 3,057 | +85% | 0 | 0 | — |
case-21 | fail→fail | 4,390 | 9,386 | +114% | 1 | 1 | 0% | 670 | 2,386 | +256% | 0 | 0 | — |
case-22 | fail→fail | 26,297 | 27,453 | +4% | 1 | 1 | 0% | 6,170 | 6,759 | +10% | 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 +50 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.