What problem does it solve?
Generic AI tools produce vague, non-compliant COMAH/Seveso III safety reports that fail regulatory review, often fabricating risk data or leaking sensitive personnel information. This Skill eliminates those risks by enforcing structured intake, strict no-fabrication rules, and mandatory de-identification before any drafting begins.
Core Features & Use Cases
- Structured Intake: Forces specificity for establishment details, tier, dangerous substances, and external QRA sources before assembling any report content, refusing to proceed on vague or missing inputs.
- Defensible Report Structure: Assembles all six required COMAH/Seveso III Safety Report elements (MAPP, safety management system, establishment description, major-accident scenarios, ALARP demonstration, internal emergency plan) and hierarchy-of-controls ranks all risk-reduction measures.
- Safety & Compliance Guardrails: Mandates de-identification of all personal and health data, records unsupplied external data as [GAP] instead of fabricating it, and requires a domain-expert SME review before any output is presented.
- Use Case: A UK upper-tier COMAH duty holder can use this Skill to structure their full Safety Report argument, ensuring all regulatory elements are covered and all quantitative risk data is traced to their external QRA provider.
Quick Start
Use the comah-safety-report-assistant skill to structure the full upper-tier COMAH Safety Report for our Grangemouth LPG terminal, using the QRA and ALARP figures from our external process safety consultant and the site details from our establishment notes.
