Show HN: JD Roast – Paste a job description, get it brutally roasted(jd-roast.openjobs-ai.com)
jd-roast.openjobs-ai.com
Show HN: JD Roast – Paste a job description, get it brutally roasted
https://jd-roast.openjobs-ai.com/
2 comments
jseabra(1)
https://jd-roast.openjobs-ai.com/
I run a recruiting AI startup, and the thing that keeps blowing my mind is how much money companies dump into sourcing tools, ATS platforms, employer branding — then turn around and publish a job description that reads like it was written by a committee in 2014.
We kept seeing the same patterns. "Competitive salary" (translation: we don't want to tell you). "Fast-paced environment" repeated four times (translation: we're disorganized). Forty-seven bullet points under Requirements (that's not a JD, that's a CVS receipt).
So I built JD Roast. You paste in a job description, it tears the thing apart across six dimensions — clarity, honesty, inclusivity, attractiveness, structure, and conversion — gives you a score out of 100, and roasts each section with specific suggestions on how to fix it.
The tone is sharp on purpose. We tried a "professional report" version first. Nobody read it. Nobody shared it. Turns out people actually engage with feedback when it stings a little.
Some stuff it catches that I think is genuinely useful: - Requirement inflation ("8 years of Kubernetes" when K8s launched in 2014) - Gendered language patterns that shrink your applicant pool without you realizing - The word "synergy." Just... no. - Missing salary info — which, depending on the state, might actually be illegal now
No login needed. No email wall on the core roast. We gate the full rewrite behind an email because yes, we're a startup and we need leads — but the roast itself and the score are completely free. Not "free trial" free. Actually free.
Some context on the tech: it's not just "throw the JD at GPT and see what comes back." Each dimension has its own detection logic — readability scoring, gendered language pattern matching, salary transparency signals, requirement inflation heuristics, etc. The LLM handles the commentary and rewrite, but the scoring rubric is deterministic.
Try it: https://jd-roast.openjobs-ai.com/
Things I'd love to hear from y'all:
1. If you've written JDs before — does the rubric make sense? Are we weighting the right stuff? 2. Is the roast tone too much, or about right? We're going for "brutally honest friend" not "mean for no reason." 3. Missing dimensions? I've been debating adding "Remote Clarity" — whether remote/hybrid/onsite is actually stated clearly. Seems important in 2026.
Happy to go deep on any technical questions or talk about the recruiting industry if anyone's curious. I spent years at Boss Zhipin in China (went from 50M to 200M users while I was there) so I have some scars and opinions on two-sided marketplaces.