What does the free prescan check before I run anything?

The prescan runs entirely in your browser with no upload and no AI. It reports word, sentence, paragraph and heading counts, and four families of flag: masculine- or feminine-coded words (the 'Gender Decoder' methodology), missing essentials (salary range, remote/hybrid/onsite policy, equal-opportunity statement, how to apply), cliché filler phrases, and an absolute degree requirement with no 'or equivalent experience' escape valve. Each family explains why it matters, and it costs nothing.

Will the review change what the role actually is?

No. The house rules require preserving the role's actual requirements and the company's voice — the model reviews for clarity, inclusivity and completeness, and every rewrite must serve those goals, never invent a different role. It never fabricates a company name, benefit, or salary figure not in your draft.

Does the review address the flags the free prescan found?

Yes. For every flag family the prescan raised, the model must either address it in the review or explain in the reconciliation notes why it intentionally left an instance as-is (some matched words are not biased in context, and some missing fields may be genuinely irrelevant).

What happens if my draft is too short to meaningfully review?

The model returns a 'too short to review meaningfully' verdict with a summary explaining what is missing, instead of inventing a full posting that is not there. A bundled example demonstrates this branch without paying for a run.

What does a run cost, and can I try it without signing in?

The prescan is free and always runs client-side. The review run is metered: a worst-case amount is reserved before the run and only what the run actually uses is charged, with both shown on the meter beside the run button. Bundled examples replay their saved results in full for free without signing in.

Job Desk — paste a rough job posting, get it reviewed

Your job posting draft

Or pick a file: it is read locally, nothing uploads until you run.

Role title — optional

Company / team context — optional

What to emphasize or fix — optional

Result

How it works

Nothing to hand? Load the — a rough posting with real bias and completeness issues and a clean review — the , where there isn't enough draft to review and the honest reply says so, or the , a posting whose real problem is an overlong, unprioritized requirements list.

Paste the posting — the prescan is free

No upload, no AI: the prescan reads the draft in your browser and reports word, sentence, paragraph and heading counts, and four families of flag — gendered-coded language, missing essentials (salary, location policy, EEO statement, how to apply), cliché phrases, an absolute degree gate. Each family explains why it matters. This part costs nothing and happens while you type.

The AI reviews — this is the metered part

A pass over the whole posting: an issue list (each with a category, a severity and the exact quote), the 6–20 highest-impact line edits each showing the original, the revision and why, a fully rewritten posting ready to publish, and a plain-English changelog. For every prescan flag, the model addresses it or explains why it left it as-is. Pricing is honest: a worst-case amount is reserved before the run and only what the run actually uses is charged — the meter next to the button shows both.

Export and iterate

Copy the rewritten posting alone, or the full report as Markdown. Download the rewritten posting as plain text, the full report as Markdown or JSON. Past reviews are saved to your SkillSafe account, so they are there on your laptop and your phone: restore one, tweak the draft, and re-run.

A derived work of @refoundai/writing-job-descriptions (github.com/refoundai/lenny-skills).