A job description grader scores a posting on clarity, bias, specificity and cliché density, then rewrites what it flags. Paste a JD to see it scored — plus a rewritten version that fixes what the score flagged. Free, no signup.
Whether a candidate can tell what they would actually do day to day. Whether requirements are genuine or inherited from the last posting. Whether the language narrows the applicant pool for reasons unrelated to the work — gendered phrasing, unnecessary degree requirements, "rockstar" and its relatives. Whether anything in it distinguishes this role from the same role at fifty other companies.
"Fast-paced environment", "wear many hats", "work hard play hard" are not neutral filler. They read as either meaningless or as a warning, and they crowd out the specifics a strong candidate is scanning for. A description that says what someone will ship in their first quarter outperforms one that lists adjectives.
Long must-have lists shrink applications disproportionately among candidates who self-assess strictly, which is a well-documented effect and part of why the same postings keep attracting the same demographics. The grader separates what is genuinely required from what was copied forward.
A better description improves who applies; it does not tell you who can do the work. Build screening criteria from the fixed version, and let AI interviews assess applicants against it.
Related: AI job description generator · Job description templates