TrustReport — {{ report.verdict.upper() }}

{% if route_data %}

{{ route_data.reason }} | Est. tokens: {{ route_data.cost.tokens }} | Est. cost: {{ route_data.cost.cost }}

{% endif %}

Confident Areas

Disputed Areas

Uncertain Areas

Models Used

0 Confidence: {{ "%.0f"|format(report.confidence) }} / 100 100

{% if trust_analysis.uncertain_count > 0 %}

⚠ 以下区域我们无法确认是否有问题

共 {{ trust_analysis.uncertain_count }} 个区域所有模型均无法确定结论。展示不确定性比展示确定性更重要——这些区域建议人工审查。

{% endif %}

Verdict: {{ report.verdict.upper() }} | Confidence: {{ "%.0f"|format(report.confidence) }}/100

{% if route_data %}

{{ route_data.reason }} | Est. tokens: {{ route_data.cost.tokens }} | Est. cost: {{ route_data.cost.cost }}

{% endif %}

{% for arb in report.arbiters[:2] %}

Brain #{{ loop.index }}

{% if arb.passed %}PASS{% else %}FAIL{% endif %}

{{ "%.0f"|format(arb.score) }}%

{% if arb.issues %}

{{ arb.issues[0].get('description', arb.issues[0])|string|truncate(150) }}

{% elif arb.suggestions %}

{% endif %}

{% endfor %}

Disagreement Details

{% for d in divergence.disagreements %}

{% if d.type == 'finding' %} [{{ d.get('severity', '?').upper() }}] {{ d.get('area', d.get('description', ''))[:80] }} {% elif d.type == 'verdict' %} Verdict Split {% elif d.type == 'score' %} Score Gap {% else %} Disagreement #{{ loop.index }} {% endif %}

{% if d.type == 'finding' %}

{% elif d.type == 'verdict' %}

{% elif d.type == 'score' %}

{% endif %}

{% endfor %}

Findings ({{ report.findings | length }})

{% for f in report.findings %} {% if trust_analysis %} {% set trust_level = trust_analysis.finding_trust.get(f.area, 'disputed') %} {% else %} {% set trust_level = 'disputed' %} {% endif %}

{{ f.severity | upper }} {% if trust_level == 'confident' %}✓ Confident {% elif trust_level == 'disputed' %}⚠ Disputed {% else %}✗ Uncertain {% endif %}

{{ f.area }} · Source: {{ f.source }}

{% if f.evidence %}

{% endif %} {% if trust_level == 'disputed' and trust_analysis and trust_analysis.dispute_details.get(f.area) %}

Model Votes on This Finding

{% for vote in trust_analysis.dispute_details[f.area].votes %}

{% if vote.attitude == 'agree' %}✓ {% elif vote.attitude == 'dispute' %}⚠ {% elif vote.attitude == 'oppose' %}✗ {% else %}– {% endif %} {{ vote.model }}

{% endfor %}

分歧点:

{% endif %}

{% endfor %}

Arbiter Votes ({{ report.arbiters | length }})

{% for a in report.arbiters %} {% endfor %}

Role	Model	Verdict	Score	Issues
{{ a.role }}	{{ a.model }}	{% if a.passed %} [PASS] {% else %} [FAIL] {% endif %}	{{ "%.0f"\|format(a.score) }}	{% if a.issues %} {% for issue in a.issues %} {{ issue.get('description', issue) \| string \| truncate(100) }} {% endfor %} {% else %} — {% endif %}

Uncertainty — What We Cannot Confirm ({{ report.uncertainty | length }})

{% for u in report.uncertainty %}

{{ u.severity | upper }} ✗ Uncertain

Reason: {{ u.reason }}
Suggestion: {{ u.suggestion }}

{% endfor %}

Blind Review (No-Context Final Review)

{% if blind_data.agreement %}CONFIRMED{% else %}DISPUTED{% endif %} {% if blind_data.agreement %}✓ Blind Confirms{% else %}⚠ Blind Disagrees{% endif %}

Model: {{ blind_data.model }} | Score: {{ "%.1f"|format(blind_data.score) }}/100

Verdict: {{ blind_data.verdict.upper() }} {% if blind_data.agreement is not none %}
Agreement with Arbitration: {% if blind_data.agreement %} ✓ CONFIRMED — Blind review independently reached the same conclusion. {% else %} ⚠ DISPUTED — Blind review reached a different conclusion. {% endif %} {% endif %}

{% if blind_data.rationale %}

Rationale: {{ blind_data.rationale }}

{% endif %} {% if blind_data.findings %}

Blind Review Findings

[{{ f.get('severity', 'medium').upper() }}] {{ f.get('description', '') }} {% if f.get('type') %} ({{ f.get('type') }}){% endif %}

{% endif %}

Cross-Family Validation

{% if report.cross_family_validated %}VALIDATED{% else %}WARNING{% endif %} {% if report.cross_family_validated %}✓ Cross-Family Validated{% else %}⚠ Same Family (Warning){% endif %}

Brain1 Family: {{ report.brain_families[0] }} | Brain2 Family: {{ report.brain_families[1] }}

{% if report.cross_family_validated %} BrainOne and BrainTwo use different provider families, ensuring optimal adversarial isolation. {% else %} BrainOne and BrainTwo use the same provider family. Cross-family adversarial isolation is not active — consider using different families for stronger safeguards. {% endif %}

Model Performance ({{ model_performance.models | length }} models)

{% for mp in model_performance.models %} {% endfor %}

Model	Current Vote	Current Score	Historical Pass Rate	Family
{{ mp.model }}	{% if mp.current_passed %} [PASS] {% else %} [FAIL] {% endif %}	{{ "%.0f"\|format(mp.current_score) }}	{% if mp.historical_pass_rate == 'N/A' %} N/A {% else %} {{ "%.0f"\|format(mp.historical_pass_rate_num * 100) }}% ({{ mp.historical_total }} reviews) {% endif %}	{{ mp.family }}

{% if model_performance.summary %}

Summary

{% for line in model_performance.summary %}

{% endfor %}

{% endif %}

Evidence Chain

SHA-256: {{ report.evidence.hash[:24] }}... Full Audit Trail

Full Hash: {{ report.evidence.hash }}
Algorithm: {{ report.evidence.algorithm }}
Timestamp: {{ report.evidence.timestamp }}
Isolation Level: {{ report.evidence.isolation_level }}
{{ k }}: {{ v }}

{% if cost %}

Audit Cost Details

{% for entry in cost.calls %} {% endfor %}

Model	Provider	Prompt Tokens	Completion Tokens	Cost (USD)
{{ entry.model }}	{{ entry.provider }}	{{ entry.prompt_tokens }}	{{ entry.completion_tokens }}	${{ "%.4f"\|format(entry.cost) }}

Total: ${{ "%.4f"|format(cost.total) }} {% if cost.comparison_total %} Full audit estimate (all top-tier): ${{ "%.4f"|format(cost.comparison_total) }} {% endif %} {% if cost.cache_hit_rate is not none %} Cache hit rate: {{ "%.0f"|format(cost.cache_hit_rate * 100) }}%, saved ~${{ "%.4f"|format(cost.cache_saved) }} {% endif %}

{% endif %}

Cross-Examine (Arsenal MVP) ({{ cross_examine_result.divergences | length }} divergences)

{% if cross_examine_result.agreement %}AGREEMENT{% else %}DIVERGENCE{% endif %} {% if cross_examine_result.agreement %}✓ Models Agree{% else %}⚠ Models Disagree{% endif %}

Models: {{ cross_examine_result.models_used[0] }} vs {{ cross_examine_result.models_used[1] }} {% if cross_examine_result.single_provider %}(Same Provider){% endif %}

{% if cross_examine_result.single_provider %}

⚠️ Same provider family — second opinion from same pipeline, not cross-family validation.

{% endif %} {% if cross_examine_result.agreement and cross_examine_result.divergences|length == 0 %}

✅ Two models reached the same conclusion on this issue.

{% elif cross_examine_result.divergences %}

⚠️ Found {{ cross_examine_result.divergences|length }} points of divergence:

Divergence Details

{% for d in cross_examine_result.divergences %}

{{ d.tag }}
Model 1: {{ d.original_sentence }}
Model 2: {{ d.second_sentence }}
Reason 1: {{ d.reason_model_1 }}
Reason 2: {{ d.reason_model_2 }}

{% endfor %}

{% endif %}

Attack Results (Arsenal MVP) ({{ attack_result.strategies | length }} strategies, {{ attack_result.triggered_count }} triggered)

{% if attack_result.triggered_count == 0 %}PASSED{% else %}VULNERABLE{% endif %} {% if attack_result.triggered_count == 0 %}✓ Attack Tests Passed{% else %}⚠ Vulnerabilities Found{% endif %}

{{ attack_result.strategies | length }} strategies tested | {{ attack_result.triggered_count }} vulnerabilities triggered

{% if attack_result.strategies %}

Attack Strategy Results

{% for s in attack_result.strategies %} {% endfor %}

Strategy	Input	Result	Triggered
{{ s.strategy_id }}	`{{ s.input_used[:60] }}...`	{{ s.execution_result[:80] }}...	{% if s.triggered %}✅ Yes{% else %}❌ No{% endif %}

{% endif %}

⚠ 以下区域我们无法确认是否有问题

Disagreement Details

Findings ({{ report.findings | length }})

Model Votes on This Finding

分歧点:

Risks ({{ report.risks | length }})

Arbiter Votes ({{ report.arbiters | length }})

Uncertainty — What We Cannot Confirm ({{ report.uncertainty | length }})

Blind Review (No-Context Final Review)

Blind Review Findings

Cross-Family Validation

Model Performance ({{ model_performance.models | length }} models)

Summary

Evidence Chain

Audit Log ({{ report.audit_log | length }} steps)

Cross-Examine (Arsenal MVP) ({{ cross_examine_result.divergences | length }} divergences)

Divergence Details

Attack Results (Arsenal MVP) ({{ attack_result.strategies | length }} strategies, {{ attack_result.triggered_count }} triggered)

Attack Strategy Results

⚠ 以下区域我们无法确认是否有问题

Disagreement Details

Share this audit

Findings ({{ report.findings | length }})

Model Votes on This Finding

分歧点:

Risks ({{ report.risks | length }})

Arbiter Votes ({{ report.arbiters | length }})

Uncertainty — What We Cannot Confirm ({{ report.uncertainty | length }})

Blind Review (No-Context Final Review)

Blind Review Findings

Cross-Family Validation

Model Performance ({{ model_performance.models | length }} models)

Summary

Evidence Chain

Audit Log ({{ report.audit_log | length }} steps)

Cross-Examine (Arsenal MVP) ({{ cross_examine_result.divergences | length }} divergences)

Divergence Details

Attack Results (Arsenal MVP) ({{ attack_result.strategies | length }} strategies, {{ attack_result.triggered_count }} triggered)

Attack Strategy Results