Reading CVs one by one is the most time-consuming thing a recruiter does that adds the least value. You're not getting paid to read CVs. You're getting paid to know which ones are worth a call. This prompt does the reading for you.
The Prompt
I'm going to paste a job spec and a set of CVs. Read all of them, then rank each candidate against the spec. > For each candidate give me: 1. Name and current role 2. A fit score out of 10 with a one-line reason 3. The two strongest points in their favour 4. The biggest gap or risk in their profile 5. Whether I should call them: Yes, Maybe, or No, with a one-line reason > Here is the job spec: [paste job spec] > Here are the CVs: [paste CV 1] --- [paste CV 2] --- [paste CV 3] > Rank them from strongest to weakest fit. No fluff.
Why It Works
The fit score forces Claude to commit to a view rather than hedge. The Yes/Maybe/No call decision is the line most prompts don't include but is the only thing that actually matters. Separating CVs with --- means Claude reads each one as a distinct document rather than one long block of text.
How to Tweak It
If the client has specific dealbreakers, add them explicitly: "The client will not consider anyone without [X]. Flag and exclude any candidate missing this before scoring."
If you're screening for a retained search and need more depth: "For the top three candidates, add a paragraph on why they'd be compelling to present to a client."
If you have more than 10 CVs, split them into batches of 5 and run the prompt twice. Ask Claude to combine and re-rank at the end.
The Point
You go from 20 CVs to a ranked shortlist in the time it used to take to read three. The ones worth calling are obvious. The ones to park are already parked. You spend your time on the calls, not the reading.