Orgmetra

Evidence-centered HRIS for the full employment lifecycle.

Orgmetra is the ContextualWisdomLab system of record for people, employment, organizations, jobs, positions, assignments, candidate-to-worker linkage, performance criteria, compensation, and evidence-backed talent decisions.

The product is intentionally federated: specialist CWL products remain independently deployable and integrate through versioned package, API, event, and adapter contracts. Orgmetra does not read another product's application tables directly.

Product loop

Job evidence
  -> Task / FJA / KSAO model
  -> SME-approved job profile
  -> Candidate evidence
  -> Structured assessment and interview
  -> Evidence-backed selection decision
  -> Employment / position / assignment
  -> Longitudinal performance outcomes
  -> Validation study
  -> Revised job and selection policy

Core bounded contexts

• People and employment
• Organization, job, position, and assignment
• Talent acquisition and candidate-worker linkage
• Performance and criterion observations
• Workforce validation and decision evidence
• Audit, provenance, and purpose-bound authorization
• CWL integration hub

CWL ecosystem boundaries

• Keyverse: identity, OIDC, SCIM, federation
• Naruon: customer-owned mail/calendar/file control plane
• Psychometrics Commons + fast-mlsirm: assessment lifecycle and psychometric computation
• TEPP: temporal, event, multilevel and multiple-membership analysis
• Semantic Data Portal: occupation/skill/ability ontology and semantic catalog
• Contextual Orchestrator: bounded, evidence-grounded AI assistance
• Clearfolio + NewsDOM: document viewing and PDF-to-DOM
• MHTML ETL Gateway + mightyETL: governed migration and CDC
• RankWeave + ThreadWeave + LineageWeave: retrieval, conversation structure and inferred evidence lineage
• Inkspan + DiagramWeave: authoring and diagrams

Non-negotiable contracts

1. person_record, employment_record, organization_unit, job_profile, position_record, and assignment_record are separate concepts.
2. Effective time and system-recorded time are preserved independently.
3. Database objects are normalized to 3NF and use descriptive two-or-more-word snake_case names.
4. Public identifiers are opaque; credentials are never HR person identifiers.
5. PII required for authorized HR work remains usable. Protection is achieved with purpose-bound authorization, least privilege, encryption, retention and audit rather than indiscriminate masking.
6. LLM output is draft evidence, never an autonomous high-impact employment decision.
7. Inferred lineage is not authoritative audit history.
8. No cross-service application-table access.

Documentation map

• docs/PRD.md
• docs/TRD.md
• ARCHITECTURE.md
• docs/USER_STORIES.md
• docs/STORYBOARD.md
• docs/WIREFRAMES.md
• docs/STORYBOOK.md
• docs/UML.md
• docs/ERD.md
• docs/DATA_MODEL.md
• docs/API_CONTRACT.md
• docs/SECURITY.md
• docs/THREAT_MODEL.md
• docs/TEST_STRATEGY.md
• docs/OPERABILITY.md
• docs/TRACEABILITY.md
• docs/adr/README.md
• docs/doctoring/REFERENCES.md

Implementation slices

• packages/orgmetra-domain: bitemporal people, employment, position, assignment, and candidate-worker invariants.
• .github/workflows/quality.yml: pinned Python 3.11-3.14 quality matrix with an aggregate required check.

Status

The foundation documentation is proposed in PR #2. The bitemporal domain kernel is proposed in a stacked implementation PR and is not protected default-branch truth until both PRs merge in dependency order.