Guide
How to set up a Claude project for recruiting
For recruiting, set up one Claude project per open role and give it three things. Instructions with the role, the hiring rule agreed before interviews, and the rule that only job-related evidence counts. Knowledge files: the role scorecard, the interview scorecard and the job post. And a brief for the job that repeats for every candidate, the summary before the debrief: each interviewer's ratings as submitted, the evidence under each competency, anything not job-related removed, disagreements and gaps flagged, and the hiring rule applied before any recommendation. Projects are on every Claude plan (up to five on Free); keep candidates' data on your company's account, and on Team and Enterprise share the project with the hiring team.
Short answer
- One project per open role.
- Instructions: the role, the hiring rule, job-related only.
- Knowledge: role scorecard, interview scorecard, job post.
- The summary brief: ratings before opinions, rule before recommendation.
What goes where
| Part of the project | What to put in it | Example |
|---|---|---|
| Instructions | The role, the hiring rule, what counts as evidence | "Customer success manager. Hire only if every must-have is 3 or higher, with no 1 and at least one 4. Only job-related evidence." |
| Knowledge | What every summary is checked against | Role scorecard, interview scorecard, job post |
| Chats | One candidate each | "Summary for candidate B before Thursday's debrief" |
From the Claude Help Center's page on projects, read 5 November 2026.
The summary brief to copy
Candidate summary brief
"Summarise these interview notes for the debrief.
First, list each interviewer's rating per competency as submitted. Then, under each competency, what the candidate said or did, close to their words, and who saw it. Put adjectives without evidence under 'Unsupported impressions'.
Remove anything not job-related (age, looks, family, health, religion, nationality, 'fit' without a definition) and write '[removed: not job-related]' with a count.
Flag any competency where ratings differ by 2 or more, and any must-have nobody assessed. List concerns with their evidence and open questions with how to answer them.
Apply the hiring rule and state the result, then give a recommendation in its own section, using only the evidence above.
Notes: [paste]"
A worked example
Illustration: a made-up business. A 40-person software company is hiring a customer success manager. Three interviewers rated candidate B on the interview scorecard right after their interviews, before talking to each other. The summary, in part:
| Competency | Ana | Ben | Chris |
|---|---|---|---|
| Handles an upset customer (must-have) | 3 | 3 | 4 |
| Explains a product change clearly (must-have) | 3 | 2 | 4 |
| Plans a renewal (must-have) | 3 | Not assessed | 3 |
| Works with the product team | 4 | 3 | 3 |
From the summary
Disagreement: "Explains a product change": 2 to 4. Ben's evidence: "Listed the new pricing tiers but didn't say what changes for the customer." Chris's: "Walked through the change from the customer's side and named the one action they need to take." Discuss first.
Removed: [removed: not job-related] × 1.
Rule: not met yet. A 2 on a must-have is a no-hire unless the debrief names evidence that resolves it.
Recommendation: resolve the product-change rating in the debrief before deciding; the other must-haves meet the bar.
The recruiter reads the summary against the original notes before the debrief, and the hiring team decides. The project keeps the scorecard and rule in one place, so every candidate's summary is checked against the same bar.
Set yours up
- Create one project per open role, on your company's account.
- Write the instructions: the role, the hiring rule, job-related evidence only.
- Add the knowledge: role scorecard, interview scorecard, job post.
- Collect ratings before anyone talks, then run the summary brief.
- Delete the project's candidate files when the role closes, as your retention policy says.
US rules on what may be asked are in the EEOC's guidance linked below; in Japan, see the Japanese version of this page.
Next step: after the offer
Once someone accepts, the same project can plan their first weeks. Onboarding new employees with AI covers the plan. For the scorecard every interviewer rates on, see one interview scorecard for the hiring team.
Skills for this
- HiringCandidate summary for a debriefSummarizes interview notes, scorecards and CV into a debrief brief: evidence per competency, disagreements, concerns, open questions, and a separate recommendation.
- HiringStructured interview scorecardBuilds a structured interview scorecard: 4 to 6 job-related competencies, 1 to 4 behavioral anchors, evidence fields and a hire/no-hire rule set in advance.
- HiringRole intake and scorecardRuns the hiring-manager intake and turns it into a role scorecard: mission, measurable outcomes, 5 or fewer must-haves, level, deal breakers, loop and sourcing plan.
- HiringReference check guide and summaryWrites a reference call guide (consent, scope, examples, the rehire question) and turns call notes into a summary with red flags.
Questions
How can recruiters use Claude?
Set up one project per open role: instructions with the role, the hiring rule and the rule that only job-related evidence counts; knowledge files with the role scorecard, the interview scorecard and the job post; and a chat per task. The job that repeats for every candidate is the summary of interviewers' notes before the debrief.
Claude or ChatGPT for recruiters?
Either works: both have projects with instructions and files, and both keep business data out of training by default on their team plans. Choose the one your company already has a business account for, because candidates' personal data shouldn't go into a personal account.
Can AI decide which candidate to hire?
No. It can organise the evidence: ratings as submitted, quotes under each competency, gaps and disagreements, and whether the agreed rule is met. The hiring team decides in the debrief.
AI skill
Have your AI prepare the hiring debrief on a candidate
Put this skill in AI Skill Pod and everyone on your team gets it in their Claude, Cursor or ChatGPT (Pro, Business, Enterprise): one invite link, free to start. Candidate summary for a debrief has your AI turn interviewers' notes into a summary the debrief can trust: ratings listed before opinions, evidence under each competency, anything not job-related removed, disagreements and gaps flagged, then the hiring rule applied before any recommendation.
14 days free · No card needed
Sources
- Claude Help Center: What are projects? — project instructions and a knowledge base per project; every plan, Free up to five projects; sharing on Team and Enterprise as Can view or Can edit (read 2026-11-05)
- Public Service Commission of Canada: Appointment processes, how to conduct interviews — board members assign independent ratings shortly after each interview; reaching consensus as a group has been found more valid; notes are a record of what the applicant said or did, not the interviewer's inferences or judgements (read 2026-11-05)
- US EEOC: Prohibited employment policies and practices — race, sex, national origin, age and religion are irrelevant to hiring decisions (read 2026-11-05)
- Anthropic Privacy Center: commercial products — Claude for Work not used for training by default (read 2026-11-05)