People Management Prover MCP Connector for Claude

You must: (1) define job-related CRITERIA — derived from essential job functions, behaviorally anchored, defined BEFORE evaluation begins. "Culture fit" is not a criterion — name the observable behavior, (2) AUDIT for bias — adverse impact analysis using the 4/5ths rule, pipeline-stage analysis, and documented mitigations. "We treat everyone equally" is not a bias audit — it is a measurement refusal, (3) verify LEGAL compliance — name the jurisdiction, the statute, the protected classes, and the prohibited inquiries. "We follow the law" without specifics is non-compliance, (4) use VALIDATED assessment methods — cite predictive validity coefficients from Schmidt & Hunter or equivalent meta-analyses. "We use interviews" is not validation — structured behavioral interviews r = 0.51, (5) design DEVELOPMENTAL feedback — behavioral, criteria-referenced, forward-looking. Feed Up + Feed Back + Feed Forward. Not "good job" and not "needs improvement." If rejected, the HR decision has a structural deficiency. Fix before concluding. Structured reflection tool for people management and recruitment reasoning — forces job-analytic criteria definition, adverse impact analysis, jurisdictional legal compliance, psychometric assessment validation, and developmental feedback design before any hiring, evaluation, or workforce decision. Grounded in I-O Psychology (Schmidt & Hunter, 1998), employment law, and Hattie's feedback research. Catches Criteria Absent (evaluating candidates without defined, job-related criteria — "she just didn't feel like the right fit." This is not assessment — it is bias cosplaying as intuition. Unstructured interviews have a predictive validity of r = 0.20 (Schmidt & Hunter, 1998). That means 96% of the variance in job performance is NOT explained by the interview. A structured behavioral interview with anchored scoring: r = 0.51. Work samples: r = 0.54. The difference: structured interviews define CRITERIA before the interview, ask the same questions to every candidate, and score responses against behavioral anchors. "Culture fit" and "gut feeling" are not criteria — they are proxies for "similar to me" bias), Bias Blind (hiring processes that produce disparate impact without detection — the 4/5ths rule (EEOC Uniform Guidelines): if the selection rate for a protected group is less than 4/5ths (80%) of the rate for the highest-scoring group, adverse impact exists. Example: 50 male applicants, 20 selected (40% rate). 30 female applicants, 6 selected (20% rate). 20% / 40% = 0.50 — well below the 0.80 threshold. Adverse impact is present. This does not necessarily mean illegal discrimination — but it DOES mean the process requires validation evidence (business necessity defense). "We hire on merit" is not a bias audit — it is a refusal to measure), Legal Non-Compliance (making employment decisions without knowing the applicable law — a US company asks candidates "When did you graduate college?" in an interview. This question is a proxy for age — a violation of the Age Discrimination in Employment Act (ADEA). A Brazilian company terminates an employee without 30-day written notice (aviso prévio) — CLT Art. 487 violation, entitles employee to one month's salary plus indemnity. A UK employer rejects a candidate because of a spent criminal conviction — Rehabilitation of Offenders Act 1974 violation. Employment law varies by jurisdiction, by protected class, and by stage of employment. "We follow the law" without naming WHICH law for WHICH jurisdiction is not compliance), Assessment Unvalidated (using assessment methods with no predictive validity evidence — "we use personality tests." Which ones? The MBTI has a test-retest reliability of 0.39-0.76 and near-zero predictive validity for job performance (Pittenger, 2005). Structured behavioral interviews: r = 0.51. Cognitive ability tests: r = 0.51. Work samples: r = 0.54. Assessment centers: r = 0.37. Graphology (handwriting analysis — still used in France): r = 0.02. If the assessment method cannot predict job performance, it is theater, not selection), and Feedback Empty (performance feedback that is vague, personality-focused, or backward-looking — "You need to be more proactive." This is a personality label, not actionable feedback. What does "proactive" look like in behavior? "In the last 3 sprint planning sessions, you waited for others to identify blockers. In the next session, I'd like you to present your risk assessment before the team discusses tasks." That is behavioral, specific, and forward-looking. Hattie: task-specific feedback d = 0.70. Praise d = 0.09. Punishment d = 0.20. Separate evaluation conversations from development conversations — combining them ensures employees hear neither). Call once per hiring decision, performance evaluation, or workforce planning action