How to Rewrite a Vague Job Description Into Clear Requirements

Anyone hiring can rewrite a vague job description into clear requirements in about 9 minutes: strip brochure fluff, force must-have versus nice-to-have bullets, and keep one measurable outcome for the first 90 days. Run it in ChatGPT or Claude before the model pads the post with “rockstar” fog.

Clear job requirements are a checkable list of skills, outcomes, and constraints, not a culture brochure and not a wishlist novel. If the model invents tools your team never uses or a degree you do not require, you never did the job. Concrete must-haves beat inspirational adjectives.

This is not a product description that is not a brochure. Product copy sells a SKU. A job post hires a person. It is not a one-paragraph bio that is not a resume. A bio introduces one human. Requirements define the seat. It is not a product spec dump into a three-line scope. Spec dumps shrink product work. This guide shrinks hiring fog into must-haves.

What you need

The vague draft JD (paste or upload). Your real must-have list in plain nouns and verbs. ChatGPT or Claude. Nine minutes and one cold read against what the team actually does this quarter.

Banned before you start: “rockstar,” “ninja,” “passionate,” inventing tools not used on the team, inventing years of experience you cannot defend, degree theater, and turning the post into a company manifesto.

Rewrite the vague post into clear requirements

1. Paste the vague draft. Name the seat in one line

Paste the full foggy JD. Add one line: “Role: [title]. Team: [name]. First 90-day outcome: [one measurable result].”

Instruct ChatGPT or Claude: “Rewrite this job description into clear requirements. Separate Must-have and Nice-to-have. Ban brochure adjectives. Ban inventing tools not listed by me. Keep the first 90-day outcome measurable. Output Role summary (3 lines max), Must-have bullets, Nice-to-have bullets, Constraints, Then stop.”

Good seat line: “Role: Support engineer. Team: Platform. First 90-day outcome: close P1 tickets under 24 hours median.” Bad: “We need a passionate self-starter who thrives in ambiguity.” Ambiguity is not a requirement you can interview against.

2. Force must-have versus nice-to-have

Shape the output as two bullet lists. Cap inventiveness at zero for tools and credentials. Every must-have must be something you would reject a candidate for missing.

Instruct: “Put only non-negotiable skills and outcomes under Must-have. Move preferences to Nice-to-have. Ban inventing certifications. Ban inventing stack items I did not name. Keep bullets under 14 words. Then stop.”

Good: “Must-have: debug Node services with logs and traces.” Bad: “Must-have: world-class communication and a growth mindset.” Mindset is not observable in the first screen.

Keep the 90-day outcome checkable

If the model writes “drive impact across the organization,” cut it. Replace with one number or one shipped artifact: tickets closed, docs published, launch shipped, error rate cut. Interviewers need a finish line, not a vibe.

3. Ban brochure words. Keep constraints honest

Tell the model: “Delete rockstar, ninja, passionate, fast-paced, wear many hats. List real constraints: location, timezone overlap, on-call, travel, contract type. Do not invent perks I did not provide.”

ChatGPT will invent “equity-heavy culture.” Claude will soften into “collaborative environment.” Gemini will invent a tool stack from a similar JD. Cut all three gifts unless you typed them.

Hard fail: any tool not on your team; any degree you do not require; any years of experience pulled from thin air; any manifesto paragraph after the requirements.

4. Cold read against the real team week

Open your calendar and the last sprint board beside the rewrite. For each must-have, ask: would we reject someone who lacks this next week? If no, move it to nice-to-have.

Read the 90-day outcome out loud. If nobody on the team can measure it, rewrite it yourself. Do not ask the model to invent a metric.

5. One human pass, then stop

Copy the surviving Role summary, Must-have, Nice-to-have, and Constraints into your ATS. Do not let the model “make it more exciting” after the fact.

Check traps: brochure adjectives, invented stack, fake years, missing 90-day outcome, culture essay. Fix by regenerating only the broken section: “Replace only Must-have. Do not invent tools. Keep bullets short.”

One pass. If the post still asks for a ninja who thrives in chaos, you never rewrote a vague job description into clear requirements. Stop.

What good looks like

Trigger: You are about to post a JD full of adjectives and no reject criteria.

Input: Vague draft, real seat line, ChatGPT or Claude, cold team read.

Output: Role summary (≤3 lines), Must-have, Nice-to-have, Constraints, one measurable 90-day outcome, zero brochure fog.

Stop: After one cold read against the real team week. No manifesto. No invented stack.

Treat the JD like a filter, not a brand film. Models love inspirational padding. Delete padding. Keep bullets scannable so busy candidates and busy interviewers share the same checklist. When models “helpfully” add trendy tools, restore only what your team runs this quarter. The cold read exists to catch invented requirements and soft adjectives. Skip that read and you will screen for theater. That wastes candidates and interview time.

Interviewers should be able to reuse the same Must-have list as a scorecard. If a bullet cannot become a yes or no in a screen, it is still fog. Keep Nice-to-have short so candidates do not treat preferences as gates. When legal or HR needs a boilerplate equal-opportunity line, place it after Constraints rather than inside Must-have. Models often bury compliance text inside skills. Move it. When you reuse this prompt next quarter, refresh the 90-day outcome first; stale outcomes create stale interviews.

Takeaways

- Paste the vague draft. Name the seat and one 90-day outcome.

- Force Must-have versus Nice-to-have. Ban brochure adjectives.

- Keep constraints honest. Do not invent tools or degrees.

- One cold team read. Then stop polishing for vibes.