Turn rough or conflicting role requirements into a source-aware scorecard, publishable job description, channel variants, and an honest release verdict.
A polished job description can still be wrong. It can hide a conflicting title, copy stale requirements, turn preferences into must-haves, invent salary or benefits, and quietly make decisions that belong to HR or legal.
This skill starts with a role contract and source register. It maps the job to observable outcomes, tests every requirement against a current business need, separates required from preferred qualifications, and holds protected-trait, compensation, classification, location, privacy, and policy decisions for the right owner. Missing facts stay missing.
The final pack includes a role scorecard, requirement decisions, canonical description, job-board variant, channel field matrix, release gates, owners, and an evidence-matched verdict. It does not screen candidates, publish a post, invent company claims, or decide what local employment law permits.
Input
The founder wants a Senior Data Engineer but lists platform ownership, analytics, ML, DevOps, support, ten tools, a degree, and ten years of Kubernetes. The reporting line and salary are not approved. Produce useful copy now without inventing decisions.
Output
A role contract with conflicts and gates, a six-outcome scorecard, requirement decisions that remove or downgrade unsupported filters, a clean canonical draft, a shorter job-board variant, and a blocked or review-ready verdict tied to the missing owner, level, and compensation decisions.
Input
A global-remote draft asks for a young native-English male, omits eligible locations and pay, requests family status in the application, and proposes automatic AI ranking. No HR, privacy, or legal owner approved it.
Provide the role brief, hiring trigger, known title and level, reporting line, employment and location details, compensation state, source material, target channels, and approval owners. Ask the agent to draft a new description, repair an existing one, reconcile conflicting versions, or create channel variants. Unknown or reserved decisions remain visible instead of being filled with plausible copy.
The rough role need, accountable hiring owner, business context, and any approved title, level, team, reporting line, or outcome information.
Known employment type, location model, eligible locations, schedule, travel, compensation state, target job boards or careers page, and publication deadline.
Existing descriptions, policy wording, brand voice, required reviews, and named HR, legal, compensation, privacy, accessibility, or publication owners. Missing evidence is allowed and stays explicit.
SKILL.md; evidence-state, requirement, inclusion, accessibility, and release controls; reusable role contract, scorecard, requirement register, canonical description, channel matrix, and release templates; OpenAI interface metadata
No reviews yet.
Output
The harmful wording is held out of publishable copy, job-related communication needs are rewritten, reserved decisions are routed to named owners, privacy-invasive instructions are removed, location and compensation gates stay open, and the release verdict remains blocked.
Input
The careers-page draft says Product Operations Lead reporting to the COO; the ATS says Product Manager reporting to Product. Responsibilities, location, employment type, and must-haves disagree. Create one controlled source and channel-ready variants.
Output
A conflict register, role scorecard, owner decisions, canonical description, job-board version, field matrix showing every controlled change, open approval items, and a verdict that cannot advance until the title, manager, scope, location, and employment type are reconciled.
Trebuh
Updated Jul 13, 2026
Automated scan passed. It found 1 finding; highest severity info.
Scanned Aug 3, 2026.
Automated scanning reduces risk but cannot guarantee a package is safe for every environment.
Creator
Ddesvar6
