Spoke

assessment-library

Every job ships with its interview kit and performance criteria — assembled from the canon, level-calibrated, and defensible (validity + adverse-impact) by construction.

Character

You lead interviewer calibration or stand up performance cycles. Today the questions and criteria are re-invented per role, inconsistent across the org, and rarely have a defensibility trail.

Problem

External. Interview questions and performance criteria are re-invented per role and inconsistent — hard to trust, hard to calibrate across levels.

Internal. You can't show that a selection process is job-related or that it passed an adverse-impact check — so it's legally exposed.

Philosophical. Every assessment item should trace to the actual work — by construction, not asserted after the fact.

Guide

assessment-library ships a paired per Family × Focus × Level: a behavioral interview bank (one STAR prompt per competency, with probes + red-flags), (concrete observable anchors at each scale point, level-calibrated — the same skill scores differently at P3 vs P6), level-calibrated performance criteria, and a validity / job-relatedness / adverse-impact (EEOC ) evidence trail. Every item is a lego brick stamped with where it came from — a canon component, a reincarnation validated item, or an operator override — so nothing is invented without a source.

Abstract

Background. Hiring and performance content is re-invented per role and rarely carries a defensibility trail, leaving selection processes legally exposed.

Methodology. A generator STAR-ifies and level-calibrates canon prose into an interview bank + BARS + performance criteria; content-validity linkage traces each competency back to canon component ids; the EEOC four-fifths rule is computed over per-group counts; validated items can fold in from reincarnation's adaptive selection when a galaxy maps to the role.

Scope. Assessment content keyed to the canon — it does NOT own the canon (job-family-agent), the psychometric items (reincarnation), the ATS workflow, or live criterion-validity studies (left null until local data confirms them).

Contribution. Paired interview + performance rubrics that are and adverse-impact-checked by construction, with full lego-brick provenance.

Evidence / Provenance. PAT-JFE-ENT-9; two real canon cells (SWE.GEN.P3, MGR.ENG.M3) run through the same generator the live path uses.

Plan

  1. 01

    Read a seeded kit

    GET /api/spokes/assessment-library/kits/SWE.GEN.P3 to see a real interview kit + BARS + criteria + defensibility verdict.

  2. 02

    Generate a kit

    POST a Family × Focus × Level to assemble (and persist) a kit from the canon.

  3. 03

    Check adverse impact

    POST per-group applicant/selected counts to /adverse-impact for the four-fifths ratio + flags.

  4. 04

    Feed development

    The same kit sharpens the career-development gap — the assess side of the learning loop.

Call to Action

Direct. curl a seeded kit and read the validity + four-fifths verdict.

Transitional. Browse the seeded kits at /assessment-library.

Spoke I/O (visual language v1)

Every toolbox spoke shares the same abstract choreography: typed inputs on the left, distilled verbs in the center, typed outputs on the right, and (when relevant) cross-spoke HTTP composition along the bottom rail. Source package: @people-analytics-toolbox/spoke-illustrations.

Assessment LibraryINPUTSMAIN ACTIONSOUTPUTSFamily × Focus × LevelProfileKeyCanon componentsCanonicalProfileValidated itemsReincarnationItem[]STAR-ify + level-calibrateBuild BARS + criteriaCheck adverse impactDefensible rubric kitRubricKit↳ validity + 4/5thsCOMPOSES WITHjob-family-agentreincarnation

Try it now

Copy this curl. Paste in any terminal. Public read — no auth needed.

assessment-library.kit.read

GET

Read a seeded rubric kit (interview bank + BARS + criteria + defensibility).

curl -sS "https://people-analytics-toolbox.vercel.app/api/spokes/assessment-library/kits/SWE.GEN.P3"

Vendor the contract

The Zod contract is the source of truth. Vendor a copy into your consumer app — you keep it; we don't break it underneath you. Re-vendor when the version bumps.

// Vendor canonical types:
// src/spokes/assessment-library/contracts/types.ts

Source path: src/spokes/assessment-library/contracts/types.ts · GitHub

Try in Dev ConsoleBrowse the seeded rubric kitsBack to all spokes

Failure

Each team writes its own questions and criteria; calibration drifts, and a challenged selection process has no job-relatedness or adverse-impact paper trail.

Success

A new role adopts the canon and instantly has a calibrated, job-related interview kit and scoring BARS — with a content-validity rationale and an adverse-impact check attached.