Auto groups spokes using fuzzy rules inspired by the consulting workbook System Documentation helper. Registry + README mining + live aggregate health (`runAggregateHealthCheck`) keep this page honest without hand-drawn diagrams. Cross-cutting primitives list every first-level module under src/lib.
Aggregate health · down · checkedAt 2026-06-03T14:05:43.515Z
Category: Operator Tool
Nested under /operator/*; composes spoke HTTP contracts + README mining + aggregate health without new schemas.
Referenced spokes (docs): worker-resolution, org-graph, manager-effectiveness, segmentation-studio
Toolbox-owned libraries that are not a single spoke — same mental bucket as the spreadsheet “system map” cross-cuts.
# data-anonymizer **Role:** cross-cutting privacy primitive. PII detection (regex catalog, header + content modes), deterministic HMAC-keyed tokenization with persistent map cache, redaction-with-spans for arbitrary text, and the toolbox's min-N privacy gate. Every other spoke that surfaces team-level rollups gates against this. **Schema:** `data_anonymize…
# org-graph — Org multi-hierarchy graph (temporal edges + overrides) — PAT-148 **Role:** Materialize HRIS-aligned org snapshots as typed, time-bounded directed edges (`reports_to`, `costs_to`, departmental / custom hierarchies); resolve ancestors with operator-override precedence; run consistency QA, snapshot diffs, dominant-dept ELT inference, substring EL…
# segmentation-studio **Role:** HRIS canonical-field normalization + multi-membership segmentation engine. Raw HRIS columns → canonical field keys with confidence; SegmentDefinitions → resolved cohorts (memberIds + segmentNodeIds). The cohort-resolution layer downstream tools (calculus, preference-modeler's by-segment cross-tab, Performix's segmentation-ada…
# worker-resolution spoke Two-Pass Self-Healing multi-source HR row resolution: learning index fortification (`Index_Reference`-style aliases) followed by deterministic merge with HRIS vault protection, supplemental fill/overwrite governance, admission policies, ladder matching, lineage, failures roster, rehire-aware termination guards.
# workforce-planning (position control + cross-system reconciliation) **Role:** Canonical **position spine** for org/location/job-family headcount, ATS requisitions, finance allocations, immutable **reconciliation** + **30/60/90 forecast** snapshots. Person-level facts stay in **segmentation-studio**; this spoke aggregates to **position grain** and aligns w…
# calculus **Role:** numerical brain. Statistical enrichment + trend analysis as service primitives that other spokes call when they need stats applied to a feed or signal. Owns the canonical `MetricEnvelope` shape — the metric × segment × period × value × provenance × enrichment object that the rest of the portfolio composes around. **Schema:** `calculus`…
# `factor-models` spoke
# forecasting **Role:** Monte Carlo simulation and value-of-information (VOI) primitives — EVPI and discrete EVSI — as **callable service APIs**, not a desktop analytics product. Owns seeded, reproducible draws; closed-form expression evaluation over named distributions; and compact decision-tree rollback for shared discrete uncertainty. **Schema:** `forec…
# metrics-catalog (HR Metrics Catalog — Catalog 1) — PAT-40 **Role:** Canonical HR metric definitions — Catalog 1 of the methodology substrate. **Donor:** `people-analyst/metric-engine-calculus` (archived) @ `97510dce446b9657cb3640c09200ad1f237f432b` (donor's `server/seed.ts` CATEGORIES + METRICS_DATA arrays; `icon` / `color` fields dropped as donor-app UI …
## performance-calibration **Owns** - Calibration impact analytics surfaced as a distinct logical service (`/api/spokes/performance-calibration/*`). - Persisted snapshots of `{ preCalibrated → postCalibrated }` tuples keyed by `{ tenant × employee × cycleId }`. **PostgreSQL** - Schema: `performance_calibration` - Tables: `heartbeat`, `calibration_pairs` …
# `performance-validity` spoke **PAT-158** ships the toolbox-native **rating instrument diagnostics** substrate: treating performance ratings as a single-item measurement and surfacing convergence, reliability/validity scorecards, predictability decomposition, trajectory diagnostics, calibration ROI probes, and change attribution narratives.
# `program-evaluation` spoke
# anycomp (Compensation Decision OS) — PAT-18 **Role:** Tenant-scoped compensation **service substrate**: comp models, market band math, employee evaluation, and cycle-level audit rows. **Donor:** `~/Vibe Coding Projects/anycomp` @ `b1a82b97f4885866ae8b029a51c32e2bf5958d70` (UI, BigQuery, Kanban, AI agent out of scope for this lift). **Status:** Live — …
## manager-effectiveness **Owner:** Toolbox — leader-level composites for Performix + ELT scorecards **Donor:** Greenfield PAT-159 (nine-domain MEI scaffolding; FiveTran-inspired default weights only) **Status:** shipped **v1.1.0** — HTTP + MCP + Drizzle-backed tenant weight profiles + empirical recalibration shim (PAT-159-FU-B)
Nested under /operator/*; composes spoke HTTP contracts + README mining + aggregate health without new schemas.
# principia-connector — TB-PRINCIPIA-01 **Role:** Read-mostly cache + MCP surface for Principia's canonical-measurement store (CanonicalVariable / Measure / Citation / Instrument / Item / Effect). **Status:** **coming-soon** — namespace reserved; Principia API isn't live yet (lands with PRN-014). The cache + MCP tool surface lands in TB-PRINCIPIA-04 once th…
# wage-benchmark spoke **Schema:** `wage_benchmark` · **Status:** live (v1) · **Contract:** `contracts/types.ts` Hourly-wage **market benchmarking** by job (SOC) × geography. Sibling to `wage-compliance`: compliance answers *"what is the legal floor?"*, benchmark answers *"what does the market pay?"*. Same audience — employers of hourly workers, who today …
# wage-compliance (Jurisdictional wage-rule engine) — PAT-79 **Role:** Tenant-aware **operational layer for jurisdictionally-variable wage compliance**: a hierarchical jurisdiction registry, a temporal rule engine with inheritance + precedence, citation-backed rule versions, and stateless single + bulk compliance evaluation against employee rosters. **Origi…
# glass-ox — PAT-GO1 **Role:** Durable run-history substrate for the Glass Ox primitive (`src/lib/glass-ox/`). The primitive stays the authoritative engine (`runStep`, the assertion catalog, `buildRunReport`, `toDataLensStage`); this spoke owns the Postgres schema, versioned wire contract, HTTP read APIs, MCP module, and the registered "plan" concept that t…
# `insight-sprint-studio`
# job-family-agent (Canonical job taxonomy + classifier + **JobFrame canon**) — PAT-71 · PAT-JF1 **Role:** Toolbox-owned canonical job taxonomy service **and the JobFrame canonical work-intelligence substrate**. Owns: (1) the SOC 2018 / O*NET 28.3 occupation registry (1,016 codes), 23 canonical job families, 26 canonical job functions, + a token-overlap heu…
# leadership-quality — does a leader own the Triple-A model? **Role:** A **0–100 Leadership Quality index** for a leader or org-unit (PAT-184, Triple-A program W5). Mike's theory: a leader is excellent to the degree they take **ownership of the Triple-A model** for their organization. A composite-over-composites — it blends component sub-indexes computed by…
# `network-analysis` (Organizational Network Analysis — ONA)
# preference-modeler **Role:** survey response collection + preference modeling. Anonymity-threshold rules that decide when a segment's responses can or cannot be surfaced. Conjoint / forced-choice / penny-allocation / pairwise comparison machinery. **Schema:** `preference_modeler` (Postgres, isolated within the shared Supabase project). **API namespace:*…
# reincarnation **Role:** adaptive psychometric diagnostic engine. Universal items (RIDs) shared across studies; study-local items (SIDs) carry virtue + IRT + reliability stats; pool-based item lifecycle (D / C / B / A / PURGATORY / INFINITY); information-gain-weighted adaptive selection. **Schema:** `reincarnation` (Postgres, isolated within the shared Su…
# `survey-orchestrator` spoke
# talent-value — the value layer of the Triple-A model **Role:** Compute the *value stack* of Mike's people-analytics model — **ELV** (Employee Lifetime Value) now; **NA% / NAV / Opportunity** layered on top in PAT-182. This is the value denominator the whole Triple-A model (Attraction · Activation · Attrition) leans on. **Donor:** greenfield (net-new compu…