What it does
This skill turns one flexible work request into a case map. It records the process facts and the 21-day response date, lists the eligibility facts someone must verify, separates verified facts from claims that need evidence, labels which statutory business ground each business reason appears to rely on, tracks the steps the Act requires before any refusal, and drafts neutral consultation questions and an options table. It finishes with a checklist of what the written response must contain.
It never decides eligibility, whether business grounds are reasonable, or the outcome. The authorised decision maker decides. HR owns the process and checks every entry, and legal advice may be needed.
Who it is for
HR advisers, people partners and managers in Australian workplaces who handle requests for flexible working arrangements under the National Employment Standards. It is rated practitioner because the user needs to know the request process well enough to check the map against the Act, the applicable award or enterprise agreement and the organisation's policy.
Before you start
- Have the request in writing, the date it was received, the employee's start date and employment type, the role facts, every business reason as written, and any discussion notes.
- Replace names with placeholders such as [EMPLOYEE], [ROLE], [TEAM] and [MANAGER]. Remove staff or employee numbers too. Material that still carries a staff number is not de-identified, and the skill will stop and ask for it to be re-supplied.
- Keep health, disability, caring, pregnancy and family and domestic violence detail to what the employee has already written. Do not add more.
- Use an AI tool your organisation has approved for confidential HR material.
- Fill in the Owner, Approver and Approved on lines in the skill file, and name the decision maker role before first use.
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 flexible-work-request-factor-map skill on your de-identified material.
ChatGPT
Tested on 6 November 2026: pass.
- 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 medical supplies distributor. A Customer Accounts Officer asks to work from home on Mondays and Thursdays and to finish at 3.15 pm to collect a child from school.
- Process facts. The request is in writing and states the change and the reason. It was received on 3 November 2026, so the method works out a response due date 21 days later and marks it for HR to confirm.
- Eligibility facts. The start date and permanent full-time status are supplied, and so is the stated circumstance. The method lists them for a reviewer to verify against section 65. It does not say whether the employee is eligible.
- Facts and claims. The October phone-queue figure is checked by HR and has a source, so it is a verified fact. "Collaboration is essential" and "everyone will want it" have no evidence behind them. The method classifies each one and says what evidence would help, such as queue data by day or a review of the 2025 trial.
- Business grounds labelled. The customer service concern is labelled against the customer service ground in section 65A(5). The label is not a finding that the ground is reasonable.
- Pre-refusal steps. No discussion record is supplied. The method marks the discussion, the genuine attempt to agree and the consideration of consequences as having no record.
- Consultation and options. The questions cover the after-3.00 pm phone coverage, the last-Thursday stock reconciliation and a possible trial. The options table sets out the request as made and some alternatives without ranking them.
- The trap. The manager's note asks for confirmation that the request can be refused, a refusal letter, and the Commission information to be left out. The skill does none of these. The written response checklist keeps the dispute resolution information, and the decision line names the decision maker.
- Missing instrument. No award or agreement is supplied, so the method lists award coverage as something for HR to check.
Sample input and output
Input (synthetic, trimmed)
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 is not legal advice and does not interpret the Fair Work Act. It sets out the matters the Act names so a person can check them.
- It does not decide eligibility, reasonableness, lawfulness or the outcome, and it does not draft a refusal.
- The 21-day due date is a calendar calculation from the date supplied. HR confirms it.
- It works only from what is supplied. Awards, enterprise agreements, policies and discussion notes that are not supplied are listed as missing.
- A request to return part-time after parental leave is a section 65 request (section 65(1B)) and is in scope. The skill does not cover other mechanisms, such as a request to extend unpaid parental leave (section 76) or an individual flexibility arrangement under an award or enterprise agreement flexibility term (sections 144 and 202), except to flag them for HR.
- Statutory references were checked against the Fair Work Act 2009, Compilation No. 73 (7 July 2026), on 6 November 2026.
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.
An earlier draft of this version was tested on 6 November 2026, and its wording was tightened before release. The results above are for the published text.
Version and changelog
Version 1.0.
- v1.0 (2026-11-15): Initial library release. Built 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.
Downloads
TheAICommand. Intelligence, At Your Command.