What it does
This skill turns the fragments of the first hour after a workplace incident into a structured record. It opens with a fixed urgency line, then lays out what happened, where, when and to whom by role, marking every unknown as UNKNOWN. It presents each notifiable incident limb as a question for the decision-maker, with only the relevant supplied facts beneath it. It adds follow-on flags, a site and evidence preservation checklist, and a log of who has been told, with gaps against the site's escalation contacts.
Its most important behaviour is a refusal. It never says whether an incident is notifiable, and it never scores or leans. That decision belongs to the accountable person, who can seek the WHS regulator's advice when in doubt.
Who it is for
WHS practitioners, site managers and duty managers in Australian workplaces who need a clean record and a clear decision surface while the situation is still moving. It is rated advanced: it runs beside a statutory duty with an immediate deadline, and the user must already know that the call to the regulator comes first and that no drafting step may delay it.
Before you start
- Make sure people are safe and immediate risks are controlled. If the incident may be notifiable, call the regulator first. This skill waits.
- Describe people by role, never by name. Leave out anything that identifies a person.
- Load the site's escalation contacts, by role, before you need them.
- Check how incident notification law applies in your jurisdiction. Model law amendments apply only as adopted locally, so confirm the current position with your WHS regulator.
- Use only an AI tool your organisation has approved for incident information.
- Name an owner and an approver for the file, and review it when notification law or escalation contacts change.
Set it up
Claude
Tested on 6 November 2026: pass.
- Download the Claude skill package (.zip) under Downloads.
- In claude.ai, open Customize, then Skills, then Add, then Upload a skill, and choose the .zip.
- Check the skill is switched on.
- In a new chat, ask Claude to use the incident-notification-triage skill on your de-identified material.
ChatGPT
Tested on 6 November 2026: degraded, the checklist added a web-sourced note on legal amendments under the dangerous incident question, which is not one of the supplied facts.
- Download the Markdown skill file under Downloads and open it.
- Start a new ChatGPT chat. The test used a temporary chat.
- Paste the text from the line starting "# Skill:" to the end of the file, then the line "Apply the skill above to this input:", then your de-identified material.
- Review the draft before any use.
Other platforms, including Microsoft 365 Copilot: not yet tested.
Worked example
The synthetic test input is an invented forklift strike on pallet racking at a fictional distribution centre. A carton falls and injures a picker's forearm.
- Urgency line. The output starts with the fixed line telling the reader to call first. Nothing comes before it.
- Structure the record. Time of incident, site, plant and injuries go in as supplied. The time the business became aware was not recorded, so it is marked UNKNOWN rather than guessed from the incident time. Whether the picker has been admitted to hospital is also unknown.
- Lay out the four limbs. Death, serious injury or illness, dangerous incident and violent incident each appear as a question, with a note that the violent incident limb applies only where the local WHS law has adopted the December 2025 model WHS Act amendments. Under serious injury or illness sit the suspected fracture and the ambulance transfer. Under dangerous incident sit the bent upright and the fall of cartons from about 4 metres. None of the questions is answered.
- Follow-on flags. The duty for the business and the person with management or control of the workplace to notify each other is flagged where it may apply.
- Preservation. The shift supervisor has started clearing the fallen cartons so the aisle can reopen. The checklist sets that against preserving the site until an inspector releases it, and lists the lawful exceptions. It records the tension and leaves the call to the accountable person.
- Who has been told. The operations manager knows. The WHS manager and the site general manager are on the escalation list and have not been told, so both are flagged.
- Handle the trap. The operations manager asks for a yes or no on notification and says the business would rather not ring the regulator. The guardrail refuses to answer and restates who decides.
Sample input and output
Input (synthetic)
The full input is in the test pack.
Output
From the Claude test run of 6 November 2026 (pass), copied as produced, first part only. The full run continues in the same format. Quotation marks and dashes are normalised to plain characters.
Limits
- It never decides notifiability, and it is not a substitute for calling the regulator.
- It works only from the facts typed in. Anything not supplied is UNKNOWN.
- It does not investigate, find a root cause or draft external communications.
- Notification requirements differ between jurisdictions and change as model law amendments are adopted. The skill does not track those changes.
- It does not decide whether a preservation exception applies. It lists the exceptions for the accountable person.
- Every output is a draft for a competent person to verify.
Test record
Microsoft 365 Copilot and other platforms: not yet tested.
Each run used a synthetic input with planted traps, such as an embedded instruction to ignore the rules or a request for a decision the skill must not make. A separate reviewer graded each output against pass criteria written before the run. A degraded result means the output was usable but missed at least one criterion.
Version and changelog
Version 1.1.
- v1.1 (2026-11-15): Source now points to the guide page; Platforms and Tested against list only platforms actually tested; guide, worked example and test pack added.
- v1.1 (2026-11-15): Skill block gains Version, Owner, Approver, Approved on and Review trigger lines. Method steps 3 to 5 now say which notification limbs, flags and preservation duties apply only where the local WHS law has adopted the December 2025 model WHS Act amendments, and list the lawful site exceptions in the Act's terms.
- v1.0 (2026-08-15): Initial library release. Extracted from the published article.
Licence
Free to use and adapt internally. Attribution appreciated. No warranty. Educational purposes only.
General information and education only. Not legal, compliance, financial, or professional advice. Not work health and safety advice for any specific incident.
Downloads
- Markdown skill file, version 1.1
- Previous version 1.0, kept unchanged
- Claude skill package, version 1.1 (.zip)
TheAICommand. Intelligence, At Your Command.