Judge — "Is the accepted result fit and sufficiently established?"

Mindset: Judge. Apply the evidence from Verify in VALUE → ASSURANCE → TRACE order. Test: "Would I defend this acceptance decision for the named purpose, harm and authority?" Verify findings: verify.md

Status and Materiality

Status vocabulary: ✅ holds · ❌ finding · ⚪ N/A does not apply with a stated reason. Every ❌ cites a Verify finding with the complete item contract. A row, criterion, discrepancy, citation, count or process record never changes the verdict by itself. Blocking requires an affected accepted claim or authority, concrete harm and material consequence. Safety/security, human acceptance authority and accepted-result identity are non-waivable subjects.

1. VALUE

Subject	Status	Evidence / finding IDs
Purpose and approved value	✅ / ❌ / ⚪	{Purpose Check below}
Domain behavior and acceptance criteria	✅ / ❌ / ⚪	{actual behavior and oracle}
Architecture and HL principles	✅ / ❌ / ⚪	{soundness evidence}
Safety and security	✅ / ❌ / ⚪	{intended/unintended behavior and harm}
Human acceptance authority and reserved effects	✅ / ❌ / ⚪	{actual ruler and boundary}

Purpose Check

Use the master HL at its Contract Baseline plus the Project North Star, never TS or a Phase HL. In one field quote the clause served and name the concrete harm at stake. A resolving but irrelevant citation fails. Green tests or TS inclusion are not sufficient.

Test three conditions:

1. Excess and adjacency — does the result deliver something the cited clause does not ask for or excludes?
2. Deferral confession — does the result ship work that its own contract assigns elsewhere?
3. Materiality — would the issue materially harm the approved value? Wording alone is not harm.

Outcome	Status	Required finding and route
Aligned	✅	quoted clause + named harm protected
Purpose failure	❌	not fit for purpose; full item contract; owner route
Reference set necessarily inconsistent	❌	contract defect; quote both clauses; owner route

Surface tension is not inconsistency. Read each clause completely. A coherent but wrong result is a purpose failure, not a contract defect.

2. ASSURANCE

Subject	Status	Evidence / finding IDs
Evidence exists	✅ / ❌ / ⚪	{RF/EV refs}
Evidence applies to accepted subject, Candidate/revision, environment, oracle/authority and dependencies	✅ / ❌ / ⚪	{subject tuples}
Evidence is sufficient for each material claim and risk	✅ / ❌ / ⚪	{selection and limits}
Permanent guards demonstrate relevant counterfactual detection	✅ / ❌ / ⚪	{guard admission; distinguish controls/diagnostics/governance}

Evidence existence and sufficiency are different judgments. A material ASSURANCE gap changes the verdict only when it leaves a material VALUE, safety, authority or accepted-result claim unestablished. If independent applicable proof already establishes the same claim, preserve the Candidate and treat the false/redundant record by its TRACE consequence.

3. TRACE

Subject	Status	Evidence / finding IDs
Governing authority and independent role lineage	✅ / ❌ / ⚪	{status/journal/dispatch}
Accepted-result identity and immutable accounting	✅ / ❌ / ⚪	{Baseline/Candidate and exact replay}
Reproducibility and citation integrity needed for material claims	✅ / ❌ / ⚪	{actual refs and methods}
Authorized continuation, item completion routes and dispositions	✅ / ❌ / ⚪	{route, owner, completion, Candidate effect}
Record-only observations	✅ / ❌ / ⚪	{finite repair or observation; no product restart}

TRACE blocks only when it materially changes authority, accepted-result identity, safety provenance, material proof or an authorized next action. Non-material wording, count, label and citation defects remain visible but receive observation or finite current-carrier repair with Candidate unchanged.

4. Finding Rulings

Copy every real Verify finding once. Preserve its class and route before deriving the verdict.

ID	Class	Subject / affected claim or authority	Fact + oracle	Harm and material consequence	Owner / completion	Route / rung	Candidate effect	Disposition
F1	VALUE / ASSURANCE / TRACE	{subject and claim}	{fact + oracle}	{harm + consequence or named absence}	{owner + observable condition}	{route}	unchanged / moves	pending / paid / promoted / not material

The highest required authority sequences only the next authorized act. It never reclassifies lower items, moves unchanged VALUE or forces every item through the highest rung. One mixed REVIEW retains one aggregate verdict.

5. Aggregate Verdict

• APPROVE — no open material item changes acceptance or the next authorized act. Visible non-material TRACE and finite record-only correction retain their own dispositions.
• REVISE — one or more material correctable items have a cited claim/authority, harm, consequence, completion condition, owner, route and Candidate effect.
• REJECT — not fit for purpose or a contract defect requires the owner route.

Verdict: {✅ APPROVE / 🔄 REVISE / ❌ REJECT}

Reason: {cite item IDs and ordered VALUE/ASSURANCE/TRACE evidence}

Contradictions with KNOWLEDGE.md

#	Knowledge item	Accepted claim	Contradiction / effect

If none: No applicable contradictions.

Checkpoint

Self-check: - [ ] Judged VALUE, then ASSURANCE, then TRACE, including every mandatory floor? - [ ] Supported every status with Verify evidence and every N/A with a reason? - [ ] Answered Purpose against Contract Baseline + North Star with a relevant clause and concrete harm? - [ ] Kept evidence existence, applicability and sufficiency distinct? - [ ] Gave every verdict-relevant item the complete finding contract, class, route and Candidate effect? - [ ] Let highest authority sequence the next act without reclassifying mixed items? - [ ] Derived exactly one verdict without using checklist, discrepancy, file, test or artifact volume as a quality objective?

Stage complete: YES / NO