Bring your own LLM key
Job descriptions that widen your pool, not narrow it
Most listings lose good candidates before they ever apply. JD Creator writes inclusive, salary-transparent descriptions and scores every draft against what actually drives applications.
No credit card. Groq has a free tier — you can be generating in about two minutes.
15
checks on every draft
each one weighted
3
providers supported
OpenAI, Gemini, Groq
0
markup on your usage
you pay them directly
The problem is rarely the role. It is how it was written.
Hiring teams rewrite the same listing four times and still get the wrong applicants. The causes are consistent, well-documented, and fixable.
- Requirement lists so long that strong candidates self-reject
- Salary hidden, so applicants opt out before they apply
- Coded language that quietly narrows and skews the pool
- Pedigree filters that screen out people who can do the job
Everything you need, nothing you have to babysit
Built as a tool you keep using, not a demo you try once.
Scored before you publish
Every draft is graded 0–100 against an evidence-backed checklist, with specific fixes ranked by how much damage each one does.
Your key, your model
Bring an OpenAI, Gemini, or Groq key. You pay the provider directly at cost — no markup, no per-seat credits, no lock-in.
Keys encrypted at rest
Keys are encrypted with AES-256-GCM and decrypted only inside the server. The interface never shows more than the last four characters.
Switch models freely
Same job description, different model. Compare quality and cost without rewriting a prompt or migrating anything.
A library, not a chat log
Every generation is saved, tagged and searchable. Copy as markdown, download, or reformat for LinkedIn in one click.
Usage you can actually see
Token counts, estimated cost and success rates per user and per model — so spend never becomes a surprise.
Three steps to your first description
- 01
Connect a provider
Paste an API key in Settings. We verify it with the provider before saving, so a bad key never reaches a generation.
- 02
Describe the role
Fill in what you know. The quality score updates as you type and tells you exactly what is weakening the listing.
- 03
Generate and publish
Pick a model and generate. Copy the markdown, download it, or switch to the LinkedIn view — already trimmed to fit.
What the score actually checks
Not a vibe rating. Each check maps to a documented cause of applicant drop-off, and every failure comes with the specific edit that fixes it.
Score your first listing- Salary range published, as a real band
- Must-have requirements kept to five or fewer
- Nice-to-haves separated so they do not read as filters
- No coded terms like “rockstar”, “ninja” or “young”
- No pedigree gatekeeping on universities or employers
- Four to seven concrete responsibilities
- Work mode and location stated plainly
- Hiring process and timeline explained
You bring the key. We never touch your bill.
Every request goes through your own provider account at their prices. Nothing is resold to you, and switching models is a dropdown, not a migration.
OpenAI
GPT-5.4 mini · GPT-5.4 nano
Google Gemini
Gemini 3.5 Flash
Groq
GPT-OSS 120B · generous free tier
Keys are encrypted with AES-256-GCM before they are stored, decrypted only inside the server, and never returned to the browser.
Questions people actually ask
What does it cost?
The app is free. You bring your own provider key and pay them directly at their prices — we add no markup and never see your bill. A typical job description costs a fraction of a cent on Groq.
Do I need an API key to start?
To generate, yes. Groq has a free tier that takes about two minutes to set up. You can sign up and explore the scoring without one.
How are my keys stored?
Encrypted with AES-256-GCM before they touch the database, decrypted only inside the server for the request that needs them. The interface only ever shows the last four characters.
Is the score just a vibe rating?
No. It is fifteen weighted checks against documented causes of applicant drop-off — salary transparency, requirement count, coded language, pedigree filters. Every failure names the specific edit that fixes it.
Can I switch models later?
Any time, from a dropdown. Nothing is tied to a provider — the same job description can be regenerated on a different model to compare quality and cost.
Who can see my job descriptions?
Only you. Each library is scoped to its owner at the database level. Administrators can see usage totals and token counts, never the content of your descriptions.
Write one that people actually apply to
Connect a key, describe the role, and see the score before you publish.