document or extract product-neutral governance finding primitives

[autholykos]

Context

rlm-tender needs governance checks similar to Pituitary's analysis outputs, but for tender artifacts:

• requirement coverage;
• evidence support;
• unsupported claims;
• stale evidence;
• contradictions against accepted company facts;
• terminology/style consistency.

Pituitary already has strong domain-specific shapes for findings, evidence, expected/observed values, remediation suggestions, content trust, and review results.

Proposed Capability

Document or extract product-neutral governance finding primitives.

Potential shared concepts:

• finding severity;
• classification;
• expected vs observed;
• evidence chain;
• source refs;
• remediation suggestion;
• confidence;
• content-trust metadata.

Non-Goals

• Do not move tender checks into Pituitary.
• Do not generalize away Pituitary's current spec/doc domain model.

Acceptance Criteria

• There is a documented mapping from Pituitary findings to generic governance concepts.
• If extraction is worthwhile, a small public package exists for shared finding/evidence primitives.
• rlm-tender can model its review report without copying large internal Pituitary structs.