People work has high-stakes writing: performance reviews, comp decisions, difficult conversations. Claude can do real work here — but only with deep context and careful editing. Below: 8 production prompts for HR teams, with explicit pitfalls and structure to avoid the categories of AI writing that hurt rather than help in people work.
Role: [NAME] Job description: [PASTE] Must-haves: [PASTE] Nice-to-haves: [PASTE] Applicant batch: [PASTE resumes/profiles] For each applicant, return: - Fit score (1-10) against must-haves - 3 specific reasons for the score - 1 question to ask in screening call to test the riskiest assumption - Recommendation (advance / borderline / pass) Do NOT use age, location, or name patterns as signals. Score on demonstrated qualifications only.
Role: [NAME] Stage of interview: [Screen / First / Final / Reference] What we still need to learn about candidates at this stage: [LIST] Design 8-10 interview questions: - 3 behavioral (past situations, STAR format) - 3 case/work-product (give them a real problem to think through) - 2 values/fit (test for what we actually value, not generic "team player") - 1 reverse (what they'd ask us) For each, include what a strong answer looks like and what a weak answer looks like.
Employee: [NAME, ROLE] Review period: [DATES] Goals at start of period: [PASTE] Measurable outcomes: [PASTE] Manager observations: [PASTE] Peer feedback: [PASTE] Draft the review: - Overall rating with 2-paragraph justification - 3 strengths with specific examples - 2-3 development areas with specific examples and growth path - Goal performance (each goal: status, what happened, what we learned) - Goals for next period (3-5, SMART) - Career conversation: where this person wants to go and what would unlock it Voice: direct, specific, no platitudes. Equal parts honest about gaps and clear about strengths.
Role: [NAME] Reports to: [WHO] Manager-stated goals for this hire: [PASTE] Team context (what's already happening): [PASTE] Build a 30/60/90: - First 30 days: learn (specific docs to read, people to meet, tools to learn) - Days 31-60: contribute (specific projects to own, smaller scope, success criteria) - Days 61-90: drive (larger ownership, leadership of a deliverable, measurable outcomes) - For each phase: 3-5 specific outcomes, the support being provided, the check-in cadence - Red flags at each phase that would indicate the hire isn't working Keep it concrete. Generic 30/60/90s don't work.
Employee: [NAME, ROLE] Current comp: [PASTE] Market benchmark (if known): [PASTE] Reason for review: [Performance / Equity / Retention / Promotion] Manager recommendation: [PASTE] Draft the comp memo: - 1-paragraph context (why this review now) - Current state vs. market - Recent performance/scope changes that justify the adjustment - Proposed new comp with breakdown (base / variable / equity) - Effective date - Communication plan (who tells them, when, message) - Risks of not making this adjustment Length: under 1 page. Voice: factual, decision-ready.
Conversation about: [PERFORMANCE / BEHAVIOR / RESTRUCTURING / OTHER] Employee: [NAME] What I need them to hear: [CORE MESSAGE] Specific examples: [PASTE] What changes from here: [PASTE] Help me prep: - Suggested opening (2-3 sentences, direct, no burying) - The 3 specific examples I should reference and how to frame each - Likely emotional responses and how to respond to each - 3 questions they'll probably ask and how to answer - Closing that's clear about next steps and timeline - What I should NOT say (defensive language, justifications, comparisons to others) Voice: respectful, specific, accountable. The goal is clarity, not punishment.
Policy change: [PASTE] Reason for the change: [PASTE] Who it affects: [PASTE] Effective date: [DATE] Questions team will probably ask: [LIST] Draft a team email: - Subject line (specific, not "Important update") - Opens with the change in 1-2 sentences (no burying) - Why (briefly, but don't over-justify) - What it means in practice (concrete examples) - What stays the same - Where to ask questions (with named person and timeline) Length: under 300 words. Voice: clear, calm, respectful.
Feedback received: [PASTE responses] Context: [What survey, what we asked, response rate] Synthesize: - 3-5 themes that came up repeatedly (with rough counts) - The most frequent single piece of feedback - Surprising or contradictory feedback (worth digging into) - What I should NOT over-react to (one person's view repeated loudly) - 3 specific actions to take based on the themes - What to communicate back to the team (and what to keep private) Be honest about themes that are uncomfortable. The whole point of anonymous feedback is to surface what people won't say in 1:1s.
These prompts work 5x better inside a properly configured Claude Project loaded with your context. See our full function guide for the setup walkthrough, or have Treetop build the Project for your team.