Screen a stack of résumés without reading all of them

The problem. You have forty résumés for one role, no recruiting team, and a product to build. Reading all forty carefully isn’t realistic, but skimming risks missing the right person.

The play. Write the rubric once — the specific things that actually matter for this role — and have Claude score the stack against it consistently, so your limited attention goes to the real shortlist instead of the full pile.

Setup (10 min). Write HIRING_BAR.md: the 3-4 things that actually matter for this specific role, and what a strong versus weak signal looks like for each. Be specific — “good communicator” isn’t a rubric line; “wrote clearly about a technical decision in their application” is. If you haven’t written the JD yet, see the job-description playbook first.

The prompt:

Here's HIRING_BAR.md for this role. Here are the résumés / applications:
[paste or point at the folder].

For each candidate, score against each item in HIRING_BAR.md as strong /
adequate / weak / unclear, with one line of evidence from their actual
application for each score — not a general impression.

Then rank all candidates and tell me the top 5-8 worth a real read. For
anyone scored "unclear" on something important, say so rather than guessing
in their favor or against them.

What you get. A shortlist you can trust was built consistently against your own bar, not vibes that drift between résumé 3 and résumé 30 the way human attention naturally does.

What to watch for.

• The rubric is doing all the work here — a vague HIRING_BAR.md produces a vague, unhelpful score no matter how careful the scoring process is.
• Read the top 5-8 yourself, fully, before deciding anything. This narrows your reading list; it doesn’t replace your judgment on who to actually interview.

Make it repeatable. Reuse and refine HIRING_BAR.md for the next role in the same function — you’ll notice which lines actually predicted a good hire and which didn’t, and the rubric gets better each time you hire.