# EM-PEO-03 Competencies and Qualifications — independent adjudication ## Verdict Complete **WM-PER-009** as the definitional root for scheme-scoped skill and competency concepts only. **Skill** and **Competency** are related but distinct definitions sharing one identity/lifecycle pattern — two typed profiles inside WM-PER-009, not aliases, not separate roots. **ProficiencyScale**: new candidate, independent versioned root, identifier unassigned. **CompetencyAssessment**: profile over WM-ACT-034, no new root. **Qualification**: reuse WM-PER-008. **Credential**: reuse WM-XCT-017, with WM-PER-013 as the professional-licence profile. Person capability holding is a further identifier-unassigned candidate; it is not WM-PER-009. No identifiers allocated. ## Evidence Frozen dossier only. WM-PER-009 is reserved (`vr.wm-per-009`, migration-boundary-review, described-previous-version) but **carries no spec in `specs`** — its content is unavailable. Its `existing_spec_ref` points at `models/activity-work/K1-function-and-capability.md`, an activity/function document, and its validation flag reads «способность человека ≠ функция системы». Its stated purpose, "What an agent is able to do", and owner, "the capable agent", describe a *holding*, not a definition. Specs supplied: WM-PER-008, WM-PER-013, WM-ACT-034, WM-XCT-017, WM-ORG-004, WM-PER-001, WM-KNW-005 — all `publishableCanonical: false`; four carry single-provider waivers. The frozen relation ledger contains only three WM-ORG rows; no competency relations exist. ## Identity/mastership Twelve identities stay distinct: (1) skill/competency concept definition; (2) scale definition and its version; (3) inter-scheme mapping assertion; (4) person capability assertion; (5) assessment event; (6) assessment result and criterion outcome; (7) self-assessment as a party-relationship variant of (5); (8) verified evidence item; (9) education qualification definition and personal award; (10) professional licence; (11) credential/attestation instance; (12) role expectation. Definitions are mastered by the owning scheme authority (ESCO, SFIA, local catalogue). Assertions are mastered by their asserter. Awards by the awarding body, credentials by the issuer, licences by the regulator, role expectations by the position owner. ## Skill/competency definitions WM-PER-009, if completed, must own: concept identity qualified by scheme plus version; preferred and alternative labels with language tags; definition text; concept type (knowledge, skill, competency, behaviour); composition edges for composite competencies; applicability and context-of-performance notes; deprecation and successor lineage; versioned crosswalks to external schemes with declared loss. It must **not** own proficiency levels as its own values, person holdings, assessment execution, qualifications, credentials, or role requirements. Skill is atomic ability; Competency is a composite of knowledge, skill and applied behaviour in a stated context, frequently carrying level descriptors — different semantics, identical governance shape, hence profiles rather than roots. Boundary against WM-KNW-005: an Agent Skill is a versioned instruction package for an AI agent; it is not a human competency and must never resolve to the same identity. ## Scales/mappings A scale is not an attribute of a competency. ESCO reuse levels, SFIA's seven responsibility levels and a local 1–5 ladder are reused across many concepts, are independently authored and are versioned on their own cadence — independent versioned root, identifier unassigned. It owns level set, ordering, anchor descriptors, measurement level (usually ordinal), and version lineage. Rescaling produces a successor version; finalised results are never recomputed against it. Mappings are addressable, versioned assertions carrying source and target versions, mapping authority, strength, direction, purpose and an explicit loss statement. Concept-to-concept mapping is owned by WM-PER-009; level-to-level mapping by the scale model. **No exact level equivalence may ever be asserted between ESCO, SFIA and a local scale**; a mapping states comparability under a stated purpose or nothing. ## Person assertions The person capability assertion — this person holds this competency at this level on this scale, per this asserter, over this interval — has independent identity and lifecycle from every definition and every assessment. It is a candidate, identifier unassigned. It references WM-PER-001 by anchor only; no competency attribute may land on the person record. Each assertion pins competency version, scale version, level, asserter, basis (self-declared, assessed, credential-derived, observed), evidence references, assurance, validity window and decay or revalidation rule. Self-declared assertions carry no verification and are always distinguishable in any projection. ## Assessment/evidence Profile WM-ACT-034 as CompetencyAssessment. It already supplies subject reference, version-pinned criteria binding, declared method and coverage, bound outcome scale and decision rule fixed before results, evidence-to-criterion linkage, criterion outcomes with indeterminate and not-applicable preserved, aggregation, conclusion with named decision authority, validity window, assessor competence and impartiality, review and moderation, and immutable finalisation. Self-assessment is the first-party value of its existing party-relationship attribute with the assessor equal to the subject and no independent review — not a separate model. Verified evidence is an evidence item with digest, source, sufficiency judgement and a second- or third-party assessor. Its declared retention gap and empty relation ledger carry into this profile unchanged. ## Qualification/credential/licence Reuse WM-PER-008 for qualification definition, award eligibility, conferral and the personal award; it already separates participation, completion, assessed achievement, definition, award, credential and recognition as non-equivalent. Reuse WM-XCT-017 for the issued credential instance: issuance, secured representation, status method and freshness, suspension, revocation, expiry, renewal, presentation and verification. WM-PER-013 profiles it for professional licences and already states that a valid proof is not claim truth and not practice authorisation. Issue, verification, validity and current competence are four separate judgements: a verification result is technical; validity is time-qualified standing; issuance is a preserved historical fact; competence is an assertion requiring its own evidence. ## Role/course boundary Role expectation lives on WM-ORG-004 (`competency-and-credential-requirements`, `screening-clearance-and-approval`) as a requirement on the vacant seat with requirement strength, never as an assertion about an occupant. Course completion lives in WM-PER-008 (`planned-component-participation-attendance-and-completion`), which already forbids reading attendance or completion as assessed achievement. The negative case fails on both models: completing a course yields a component-completion record, no expert-level assertion, and no credential. ## Time/provenance/privacy Keep learning, assessment, decision, conferral, issue, revocation, verification, observation and ingestion times distinct, RFC 3339 with explicit offset. Every assertion names asserter, method and evidence. Competency and assessment data about an identified person is personal data and may be special-category where health, disability or accommodation is implicated; disclosure is minimum-necessary and purpose-bound, self-declared values are always marked, and superseded assertions are retained rather than overwritten. ## Governance Owner: HR lead. Candidate master systems HRIS, personnel register, ATS, LMS are masters of *assertions and records*, never of scheme definitions. Scheme authorities master their own concepts and scales. Awarding bodies, issuers and regulators master awards, credentials and licences respectively. ## Acceptance scenario Competency "secure code review". Assessment A: manager review, local 1–5 scale v2, level 4, single assessor, no moderation. Assessment B: work-sample against SFIA-aligned criteria, third-party assessor, evidence-linked, level 4 on the SFIA scale. Both finalised, both retained; neither supersedes the other. The local-to-SFIA mapping asserts partial comparability for development-planning purposes only and declares loss — no equivalence of the two level-4 values is derivable. A holds an expired certificate for the same competency: the issuance fact and award remain true, credential status is expired, and no current-competence assertion follows from it. Scale v2 → v3 rescaling produces a successor scale; both results keep their original bindings. ## Invariants 1. Definition is not holding. 2. Concept identity is scheme-plus-version qualified. 3. Level is meaningless without its scale version. 4. Every assessment states scale, method, assessor and date. 5. Finalised results are never recomputed against a later scale. 6. Self-declared and verified assertions are always distinguishable. 7. Mapping states loss; exact level equivalence is prohibited. 8. Course completion never implies proficiency. 9. Certificate expiry does not erase the issuance fact. 10. Credential validity is not current competence. 11. Role expectation is not a person assertion. 12. Human competency and Agent Skill never share identity. 13. Missing assessment is unknown, not zero. ## Minimal model set WM-PER-009 (complete: definitions only) · ProficiencyScale (new candidate, unassigned) · PersonCapabilityAssertion (new candidate, unassigned) · WM-ACT-034 profile · WM-PER-008 · WM-XCT-017 + WM-PER-013 · WM-ORG-004 · WM-PER-001 reference. ## Holds WM-PER-009 has no specification in this dossier; its purpose, owner and legacy K1 spec reference conflate person capability with system function and must be resolved before completion. All supplied bases are non-canonical reviewable drafts, four under single-provider waivers. No approved relation rows exist for any competency edge; every link above is a proposal. Two candidates lack registry allocation and independent evidence. ESCO, SFIA, CTDL, CLR, Open Badges and VC crosswalks, scale registries, fixtures and source pins are unverified. No claim of canonical completeness, installability or publication readiness is made.