Assess knowledge-layer integrity

Core meaning

Capability purpose: Keep owner knowledge navigable, source-linked, current, and safe by producing inspectable maintenance findings and follow-up work.
Meaningful outcome: An assessment identifies the checked scope, rules/profile, evidence, defects, severity, affected objects, proposed responses, review tasks, and unresolved limitations.
Boundaries — includes: scope manifest, broken/orphan links, missing summaries or indexes, stale/duplicate concepts, missing or weak source references, unresolved contradictions, unconfirmed labels, visibility/sensitivity problems, review status, severity, and attributable findings.
Boundaries — excludes: silently editing knowledge, deciding every proposal, merging duplicates, deleting sources, changing visibility, rebuilding indexes, and performing generic software-code linting.
Terms and concepts: A finding may be represented as a schema.org/Review, schema.org/Action, or structured quality record linked to the affected schema.org/CreativeWork or knowledge object.

Actor: The owner, an authorized knowledge-maintenance process, or a review agent.
Command / intent: Check a declared knowledge scope against a named integrity and safety profile.
Current state: The target objects, links/indexes, source references, lifecycle states, visibility/sensitivity metadata, assessment profile/version, and authority can be determined.
Policies / invariants: Findings preserve evidence and checked-time context; inferred defects remain distinguishable from confirmed errors; material changes require their own capability and authority; important knowledge is not silently rewritten; unavailable checks and coverage gaps remain visible; generated indexes are treated as rebuildable projections.
Transition: Evaluate the bounded scope, record findings and severity, link affected objects and evidence, and create proposals or review tasks where appropriate.
Result: A completed, partial, failed, or blocked integrity assessment with actionable findings and coverage evidence.
Events / effects: May create review tasks, owner-attention items, proposals, source-link corrections, index rebuild requests, or concept-lifecycle evaluations through separate capabilities.
Unknowns: Universal lint profiles, severity scales, scheduling, false-positive handling and auto-fix boundaries are not established.

Integrity assessment must not silently rewrite, delete, merge, promote, or expose knowledge.
Each finding must identify its checked scope, rule/profile, evidence and assessment time.

Produce flags, proposals and review tasks before applying material knowledge changes.

Which low-risk mechanical corrections may be standing-authorized requires a separate governance profile.

Statement	Semantic class	Evidence status	Source / note
Knowledge integrity assessment produces evidence-linked structural, evidence, meaning and safety findings while material repair remains separately governed.	rule/invariant	sourced	S11–S15.