What problem does it solve?
Generic AI-generated safety paperwork is often vague, relies exclusively on low-order controls like PPE, lacks clear action owners, leaks personal health data, and is frequently rejected by regulators. This skill eliminates those failures by enforcing a structured intake process to capture all required incident and jurisdiction details, applying jurisdiction-specific recordability/reportability tests, mandating strict de-identification of sensitive data, and requiring domain-expert review before any output is delivered.
Core Features & Use Cases
- Jurisdiction-specific compliance testing: Applies the correct recordability/reportability criteria for US OSHA 29 CFR 1904 (Forms 300/300A/301), UK RIDDOR 2013, EU member-state equivalents, and routes India returns to the hse-india skill suite with mandatory state detection.
- Mandatory de-identification: Scrubs all personal identifiers, health data, and small-cell aggregated data (fewer than 5 individuals) from circulated copies, with separate access-controlled storage for legally required named submissions and re-identification keys.
- Structured intake guardrails: Refuses to provide a recordable/reportable yes/no answer until all required incident facts, jurisdiction, and return type details are captured, preventing guesswork and non-compliant determinations.
- Use Case: A safety manager can use this skill to confirm if a first-aid-only workplace laceration is recordable on the OSHA 300 log, prepare a RIDDOR specified injury report with the correct 15-day filing deadline, or generate a de-identified 300A annual summary for their team without exposing employee identifiers.
Quick Start
Ask the regulatory-returns skill to determine if your recent workplace incident is recordable under OSHA 1904 or reportable under UK RIDDOR, and generate a de-identified statutory return with the correct form and exact filing deadline.