Data-layer swarm · Governance-Decision-Router Agent · Routing-audit-trail skill · Build pillar · Published July 12, 2026

How to build routing audit trails for AI-output governance

This guide explains how to architect the routing-audit-trail skill on the governance-decision-router agent end-to-end at multi-model multi-regime governance scale: per-portfolio per-banner per-location per-decision per-canonical-routing-capture-spec + per-canonical-policy-engine-trace-spec + per-canonical-prompt-completion-snapshot-spec + per-canonical-reviewer-sign-off-spec + per-canonical-multi-LLM-pre-publish-check + per-canonical-FBC-feedback-loop + per-canonical-audit-trail + per-portfolio audit-trail.

Engage the governance-decision-router agent See the 32-agent swarm

What you will build

Per-portfolio per-banner per-location per-decision per-canonical-routing-capture-spec across per-route + per-model (per-GPT-4o + per-Claude-Sonnet + per-Gemini-Pro + per-Mistral-Large + per-Llama-3 + per-Cohere-Command + per-Anthropic-Bedrock + per-OpenAI-Azure + per-fine-tuned-model + per-RAG-retrieval-augmented + per-canonical-10-model) + per-prompt-template-version + per-system-message-snapshot + per-tool-call-trace + per-RAG-retrieval-snapshot + per-vector-store-pointer + per-embedding-model-pointer + per-temperature + per-top-p + per-max-tokens + per-seed + per-completion-snapshot + per-tokens-in-out + per-cost + per-latency-p50-p95-p99 + per-confidence-tier + per-explainability.
Per-canonical-policy-engine-trace-spec — per-OPA-Rego + per-AWS-Cedar + per-Casbin + per-Cerbos + per-Oso + per-policy-version-pointer + per-policy-decision-log-allow-deny + per-policy-explainability + per-policy-circuit-breaker-state + per-policy-fallback-rule + per-policy-handoff-to-versioned-history-regulatory-defense.
Per-canonical-prompt-completion-snapshot-spec + per-reviewer-sign-off-spec — per-SHA-256-prompt-fingerprint + per-completion-fingerprint + per-bitemporal-storage-handoff + per-immutable-WORM + per-CCPA-CPRA-DSAR-tagging + per-GDPR-DPIA-tagging + per-PII-detection-redaction (name + email + phone + SSN + credit card) + per-PHI-detection-redaction (HIPAA PHI) + per-tokens-snapshot + per-cost-snapshot + per-confidence-snapshot + per-6-reviewer-role (marketing + legal + compliance + security + clinical + financial) + per-reviewer-authentication + per-reviewer-authorization-OPA-Cedar-Casbin-Cerbos-Oso + per-sign-off-state (pending + approved + rejected + escalated) + per-reviewer-comment + per-reviewer-SLA + per-reviewer-audit-trail + per-PagerDuty-Opsgenie-incident-handoff + per-Slack-Teams-channel-handoff.
Per-canonical-multi-LLM-pre-publish-check — per-multi-LLM-ensemble + per-routing-decision-cross-check + per-policy-decision-cross-check + per-reviewer-decision-cross-check + per-LLM-confidence-scoring + per-LLM-self-consistency + per-LLM-chain-of-thought.
Per-canonical-FBC-feedback + per-audit-trail + per-cross-skill-handoff — per-realized-vs-projected-outcome + per-reviewer-approval-vs-LLM-approval + per-policy-decision-vs-LLM-decision + per-routing-decision-pattern-learning + per-policy-rule-drift-detection + per-prompt-template-drift-detection + per-model-drift-detection + per-completion-drift-detection + per-cost-vs-budget + per-latency-vs-SLO + per-reviewer-SLA-vs-realized + per-policy-circuit-breaker-recalibration + per-emerging-failure-mode-detection + per-per-decision-canonical-audit-record + per-EU-AI-Act-Article-13 + per-EU-AI-Act-Article-14 + per-EU-AI-Act-Article-15 + per-NIST-AI-RMF + per-NIST-SP-800-218A + per-NIST-SP-800-53 + per-SOC-2-Type-II + per-ISO-27001 + per-ISO-42001-AI-management-system + per-HIPAA-PHI + per-PCI-DSS + per-FedRAMP + per-FTC-substantiation-doctrine + per-FTC-MARS + per-FTC-AI-disclosure + per-FDA-AI-ML-SaMD + per-FINRA-2210 + per-CFPB-UDAAP + per-CCPA-CPRA-DSAR + per-GDPR-DPIA + per-LGPD-ANPD + per-DPDP-India + per-PIPEDA-OPC + per-CASL-CRTC + per-FDD-Item-12-Item-17-Item-19 + per-immutable-WORM + per-7-year-IRS-tax-retention + per-handoff-to-11-sibling-skills.

Why per-vendor-OpenAI-Logs-account-flat-log breaks at multi-model multi-regime governance scale

Per-vendor-OpenAI-Logs-canonical-account-flat-log ships per-account per-completion per-flat-log primitive. Per-vendor-Anthropic-Console + Google-Cloud-Vertex-AI-Logging + Azure-OpenAI-Service-Logs + LangSmith + Weights-and-Biases + Arize + WhyLabs + Helicone + Langfuse + Lakera-Guard + Robust-Intelligence + HiddenLayer + CalypsoAI-canonical-account-flat-log ship per-vendor per-native account-flat-log primitives.

At 1-account-1-model-flat-log scale per-account per-completion per-flat-log primitive is enough. At multi-model multi-regime governance scale per-per-decision-per-model-routing-capture-spec + per-per-decision-policy-engine-trace-spec + per-per-decision-prompt-completion-snapshot-spec + per-per-decision-reviewer-sign-off-spec + per-per-decision-multi-LLM-pre-publish-check + per-per-decision-FBC-feedback-loop + per-per-decision-audit-trail.

Per-account-flat-log + per-per-decision-per-model-routing-capture-blind + per-per-decision-policy-engine-trace-blind + per-per-decision-prompt-completion-snapshot-blind + per-per-decision-reviewer-sign-off-blind + per-per-decision-multi-LLM-pre-publish-blind + per-per-decision-FBC-feedback-loop-blind + per-multi-format-audit-trail-blind + per-EU-AI-Act-Article-13-Article-14-Article-15-blind + per-NIST-AI-RMF-blind + per-NIST-SP-800-218A-blind + per-NIST-SP-800-53-blind + per-SOC-2-Type-II-blind + per-ISO-27001-blind + per-ISO-42001-AI-management-system-blind + per-HIPAA-PHI-blind + per-PCI-DSS-blind + per-FedRAMP-blind + per-FTC-substantiation-doctrine-blind + per-FTC-MARS-blind + per-FTC-AI-disclosure-blind + per-FDA-AI-ML-SaMD-blind + per-FINRA-2210-blind + per-CFPB-UDAAP-blind + per-FDD-Item-12-Item-17-Item-19-blind + per-7-year-IRS-tax-retention-blind.

The operator-side architecture above per-vendor-flat-log primitive is canonical-per-decision-per-model-routing-capture-spec + per-per-decision-policy-engine-trace-spec + per-per-decision-prompt-completion-snapshot-spec + per-per-decision-reviewer-sign-off-spec + per-per-decision-multi-LLM-pre-publish-check + per-per-decision-FBC-feedback-loop + per-per-decision-audit-trail + per-portfolio-audit-trail.

What is in market today

Per-platform per-LLM-observability-vendor

OpenAI Logs, Anthropic Console, Google Cloud Vertex AI Logging, Azure OpenAI Service Logs, LangSmith, Weights & Biases, Arize, WhyLabs, Helicone, Langfuse, Datadog LLM Observability, New Relic AI Monitoring, Splunk Observability for LLM. Per-account per-completion per-flat-log primitive. Per-canonical-per-decision-per-model-routing-capture-canonical-per-decision-policy-engine-trace-canonical-per-decision-prompt-completion-snapshot-canonical-per-decision-reviewer-sign-off-canonical-per-decision-multi-LLM-pre-publish-canonical-per-decision-FBC-feedback-canonical-per-decision-audit-trail is not the primitive.

Per-platform per-AI-guardrail-vendor

Lakera Guard, Robust Intelligence, HiddenLayer, CalypsoAI, Protect AI, Aporia, Patronus AI, Guardrails AI, NeMo Guardrails (NVIDIA), Microsoft Presidio, AWS Bedrock Guardrails, Google Cloud Model Armor. Per-account per-rule-set-flat-guardrail primitive. Per-canonical-per-decision-OPA-Rego-policy-trace-canonical-per-decision-AWS-Cedar-policy-trace-canonical-per-decision-Casbin-policy-trace-canonical-per-decision-Cerbos-policy-trace-canonical-per-decision-Oso-policy-trace-canonical-per-decision-policy-version-pointer-canonical-per-decision-policy-decision-log-allow-deny-canonical-per-decision-policy-explainability-canonical-per-decision-policy-circuit-breaker-state-canonical-per-decision-policy-fallback-rule is not the primitive.

Per-platform per-PII-PHI-redaction-vendor

Microsoft Presidio, Google Cloud DLP, AWS Comprehend PII, Skyflow, Private AI, Tonic.ai, Datafleet, Privya, BigID, Securiti, OneTrust DataDiscovery, Spirion. Per-account per-rule-set-flat-redaction primitive. Per-canonical-per-decision-prompt-fingerprint-canonical-per-decision-completion-fingerprint-canonical-per-decision-prompt-completion-bitemporal-storage-handoff-canonical-per-decision-prompt-completion-immutable-WORM-storage-canonical-per-decision-prompt-completion-CCPA-CPRA-DSAR-tagging-canonical-per-decision-prompt-completion-GDPR-DPIA-tagging-canonical-per-decision-prompt-completion-PII-detection-redaction-canonical-per-decision-prompt-completion-PHI-detection-redaction-canonical-per-decision-prompt-completion-tokens-snapshot-canonical-per-decision-prompt-completion-cost-snapshot-canonical-per-decision-prompt-completion-confidence-snapshot is not the primitive.

Per-platform per-AI-governance-platform-vendor

Credo AI, Holistic AI, Truera, FairlyAI, Monitaur, Modulos, IBM Watson OpenScale, Salesforce Einstein Trust Layer, ServiceNow AI Governance, Anthropic Console Governance, OpenAI Evals. Per-account per-flat-governance-report primitive. Per-canonical-per-decision-reviewer-role-canonical-per-decision-reviewer-authentication-canonical-per-decision-reviewer-authorization-policy-as-code-OPA-Cedar-Casbin-Cerbos-Oso-canonical-per-decision-reviewer-sign-off-state-canonical-per-decision-reviewer-comment-snapshot-canonical-per-decision-reviewer-SLA-canonical-per-decision-reviewer-audit-trail-canonical-per-decision-reviewer-handoff-to-PagerDuty-Opsgenie-incident-canonical-per-decision-reviewer-handoff-to-Slack-Teams-channel is not the primitive.

How the architecture is built

Per-portfolio per-banner per-location per-decision per-canonical-routing-capture-substrate. Per-route + per-10-canonical-model + per-prompt-template-version + per-system-message-snapshot + per-tool-call-trace + per-RAG-retrieval-snapshot + per-vector-store-pointer + per-embedding-model-pointer + per-temperature + per-top-p + per-max-tokens + per-seed + per-completion-snapshot + per-tokens-in-out + per-cost + per-latency-p50-p95-p99 + per-confidence-tier + per-explainability canonical-routing.
Per-portfolio per-canonical-policy-engine-trace-spec. Per-OPA-Rego + per-AWS-Cedar + per-Casbin + per-Cerbos + per-Oso + per-policy-version-pointer + per-policy-decision-log + per-policy-explainability + per-policy-circuit-breaker-state + per-policy-fallback-rule canonical-policy.
Per-portfolio per-canonical-prompt-completion-snapshot-spec. Per-SHA-256-prompt-fingerprint + per-completion-fingerprint + per-bitemporal-storage-handoff + per-immutable-WORM + per-CCPA-CPRA-DSAR-tagging + per-GDPR-DPIA-tagging + per-PII-detection-redaction + per-PHI-detection-redaction + per-tokens-snapshot + per-cost-snapshot + per-confidence-snapshot canonical-snapshot.
Per-portfolio per-canonical-reviewer-sign-off-spec. Per-6-reviewer-role + per-reviewer-authentication + per-reviewer-authorization-OPA-Cedar-Casbin-Cerbos-Oso + per-sign-off-state + per-reviewer-comment + per-reviewer-SLA + per-reviewer-audit-trail + per-PagerDuty-Opsgenie-incident-handoff + per-Slack-Teams-channel-handoff canonical-reviewer.
Per-portfolio per-canonical-multi-LLM-pre-publish-check. Per-multi-LLM-ensemble + per-routing-decision-cross-check + per-policy-decision-cross-check + per-reviewer-decision-cross-check + per-LLM-confidence + per-LLM-self-consistency + per-LLM-chain-of-thought canonical-LLM.
Per-portfolio per-canonical-FBC-feedback-loop. Per-realized-vs-projected-outcome + per-reviewer-approval-vs-LLM-approval + per-policy-decision-vs-LLM-decision + per-routing-decision-pattern-learning + per-policy-rule-drift-detection + per-prompt-template-drift-detection + per-model-drift-detection + per-completion-drift-detection + per-cost-vs-budget + per-latency-vs-SLO + per-reviewer-SLA-vs-realized + per-policy-circuit-breaker-recalibration + per-emerging-failure-mode-detection canonical-FBC.
Per-portfolio per-canonical-audit-trail + per-portfolio-audit-trail. Per-per-decision-canonical-audit-record + per-EU-AI-Act-Article-13-Article-14-Article-15 + per-NIST-AI-RMF + per-NIST-SP-800-218A + per-NIST-SP-800-53 + per-SOC-2-Type-II + per-ISO-27001 + per-ISO-42001-AI-management-system + per-HIPAA-PHI + per-PCI-DSS + per-FedRAMP + per-FTC-substantiation-doctrine + per-FTC-MARS + per-FTC-AI-disclosure + per-FDA-AI-ML-SaMD + per-FINRA-2210 + per-CFPB-UDAAP + per-CCPA-CPRA-DSAR + per-GDPR-DPIA + per-LGPD-ANPD + per-DPDP-India + per-PIPEDA-OPC + per-CASL-CRTC + per-FDD-Item-12-Item-17-Item-19 + per-immutable-WORM + per-7-year-IRS-tax-retention canonical-audit.
Per-portfolio per-canonical-cross-skill-handoff. Per-handoff-to-11-sibling-skills canonical-handoff.
Per-portfolio per-governance-decision-router-agent-canonical-bundle. Per-routing-audit-trail + per-audit-trail-software + per-ai-routing-decision-audit-trail + per-policy-as-code-OPA-Rego-AWS-Cedar-Casbin-Cerbos-Oso + per-prompt-template-version-management + per-model-version-management + per-reviewer-workflow + per-PII-PHI-redaction + per-vector-store-version-management + per-RAG-retrieval-trace canonical-bundle.
Per-portfolio per-canonical-end-to-end-SLA. Per-routing-capture-to-policy-trace-to-prompt-completion-snapshot-to-reviewer-sign-off-to-pre-publish-to-present-to-FBC-SLA canonical-end-to-end-SLA.
Per-portfolio per-canonical-end-to-end-replay. Per-replay-spec + per-replay-trace + per-replay-decision + per-per-decision-replay canonical-replay.

Frequently asked questions

What is routing audit trail for AI-output governance at multi-model multi-regime scale?

Routing audit trail runs per-portfolio per-banner per-location per-decision per-canonical-per-decision-per-model-routing-capture-spec + per-canonical-per-decision-policy-engine-trace-spec + per-canonical-per-decision-prompt-completion-snapshot-spec + per-canonical-per-decision-reviewer-sign-off-spec + per-canonical-per-decision-multi-LLM-pre-publish-check + per-canonical-per-decision-FBC-feedback-loop + per-canonical-per-decision-audit-trail + per-portfolio audit-trail. Per-canonical-per-decision-routing-capture covers per-route + per-model (per-GPT-4o + per-Claude-Sonnet + per-Gemini-Pro + per-Mistral + per-Llama + per-Cohere + per-Anthropic-Bedrock + per-OpenAI-Azure + per-canonical-model) + per-prompt-template-version + per-completion + per-tokens-in-out + per-cost + per-latency + per-canonical-routing-field.

Why does per-vendor-OpenAI-Logs-canonical-account-flat-log break at multi-model multi-regime governance scale?

Per-vendor-OpenAI-Logs-canonical-account-flat-log ships per-account per-completion per-flat-log primitive. Per-vendor-Anthropic-Console-canonical + per-Google-Cloud-Vertex-AI-Logging-canonical + per-Azure-OpenAI-Service-Logs-canonical + per-LangSmith-canonical + per-Weights-and-Biases-canonical + per-Arize-canonical + per-WhyLabs-canonical + per-Helicone-canonical + per-Langfuse-canonical + per-Lakera-Guard-canonical + per-Robust-Intelligence-canonical + per-HiddenLayer-canonical + per-CalypsoAI-canonical-account-flat-log ship per-vendor per-native account-flat-log primitives. At 1-account-1-model-flat-log scale per-account per-completion per-flat-log primitive is enough. At multi-model multi-regime governance scale per-canonical-per-decision-per-model-routing-capture-spec + per-canonical-per-decision-policy-engine-trace-spec + per-canonical-per-decision-prompt-completion-snapshot-spec + per-canonical-per-decision-reviewer-sign-off-spec + per-canonical-per-decision-multi-LLM-pre-publish-check + per-canonical-per-decision-FBC-feedback-loop + per-canonical-per-decision-audit-trail.

How does per-decision per-model-routing-capture + per-policy-engine-trace work?

Per-portfolio per-banner per-location per-decision per-canonical-routing-capture-spec runs per-portfolio per-canonical-per-decision-route-pointer + per-canonical-per-decision-model-pointer (per-GPT-4o + per-Claude-Sonnet + per-Gemini-Pro + per-Mistral-Large + per-Llama-3 + per-Cohere-Command + per-Anthropic-Bedrock + per-OpenAI-Azure + per-fine-tuned-model + per-RAG-retrieval-augmented + per-canonical-model) + per-canonical-per-decision-prompt-template-version + per-canonical-per-decision-system-message-snapshot + per-canonical-per-decision-tool-call-trace + per-canonical-per-decision-RAG-retrieval-snapshot + per-canonical-per-decision-vector-store-pointer + per-canonical-per-decision-embedding-model-pointer + per-canonical-per-decision-temperature-spec + per-canonical-per-decision-top-p-spec + per-canonical-per-decision-max-tokens-spec + per-canonical-per-decision-seed-spec + per-canonical-per-decision-completion-snapshot + per-canonical-per-decision-tokens-in-out-spec + per-canonical-per-decision-cost-spec + per-canonical-per-decision-latency-p50-p95-p99 + per-canonical-per-decision-confidence-tier + per-canonical-per-decision-explainability. Per-canonical-per-decision-policy-engine-trace-spec runs per-portfolio per-canonical-per-decision-OPA-Rego-policy-trace + per-canonical-per-decision-AWS-Cedar-policy-trace + per-canonical-per-decision-Casbin-policy-trace + per-canonical-per-decision-Cerbos-policy-trace + per-canonical-per-decision-Oso-policy-trace + per-canonical-per-decision-policy-version-pointer + per-canonical-per-decision-policy-decision-log-allow-deny + per-canonical-per-decision-policy-explainability + per-canonical-per-decision-policy-circuit-breaker-state + per-canonical-per-decision-policy-fallback-rule + per-canonical-per-decision-policy-handoff-to-versioned-history-regulatory-defense (sibling build-pillar at /how-to-build-versioned-history-regulatory-defense-for-multi-location-operators).

What does per-decision prompt-completion-snapshot + per-decision reviewer-sign-off + per-decision multi-LLM-pre-publish do?

Per-portfolio per-banner per-location per-decision per-canonical-prompt-completion-snapshot-spec runs per-portfolio per-canonical-per-decision-prompt-fingerprint (per-SHA-256 + per-prompt-version-pointer + per-canonical-fingerprint) + per-canonical-per-decision-completion-fingerprint + per-canonical-per-decision-prompt-completion-bitemporal-storage-handoff-to-versioned-history-regulatory-defense + per-canonical-per-decision-prompt-completion-immutable-WORM-storage + per-canonical-per-decision-prompt-completion-CCPA-CPRA-DSAR-tagging + per-canonical-per-decision-prompt-completion-GDPR-DPIA-tagging + per-canonical-per-decision-prompt-completion-PII-detection-redaction (per-name + per-email + per-phone + per-SSN + per-credit-card + per-canonical-PII-type) + per-canonical-per-decision-prompt-completion-PHI-detection-redaction (per-HIPAA-PHI + per-canonical-PHI-type) + per-canonical-per-decision-prompt-completion-tokens-snapshot + per-canonical-per-decision-prompt-completion-cost-snapshot + per-canonical-per-decision-prompt-completion-confidence-snapshot. Per-canonical-per-decision-reviewer-sign-off-spec runs per-portfolio per-canonical-per-decision-reviewer-role-spec (per-marketing-reviewer + per-legal-reviewer + per-compliance-reviewer + per-security-reviewer + per-clinical-reviewer + per-financial-reviewer + per-canonical-reviewer-role) + per-canonical-per-decision-reviewer-authentication + per-canonical-per-decision-reviewer-authorization-policy-as-code-OPA-Rego-AWS-Cedar-Casbin-Cerbos-Oso + per-canonical-per-decision-reviewer-sign-off-state (per-pending + per-approved + per-rejected + per-escalated + per-canonical-sign-off-state) + per-canonical-per-decision-reviewer-comment-snapshot + per-canonical-per-decision-reviewer-SLA-spec + per-canonical-per-decision-reviewer-audit-trail + per-canonical-per-decision-reviewer-handoff-to-PagerDuty-Opsgenie-incident + per-canonical-per-decision-reviewer-handoff-to-Slack-Teams-channel. Per-canonical-per-decision-multi-LLM-pre-publish-check runs per-portfolio per-canonical-multi-LLM-ensemble (per-GPT-4o + per-Claude-Sonnet + per-Gemini-Pro per-canonical-LLM) + per-canonical-per-decision-LLM-routing-decision-cross-check + per-canonical-per-decision-LLM-policy-decision-cross-check + per-canonical-per-decision-LLM-reviewer-decision-cross-check + per-canonical-per-decision-LLM-confidence-scoring + per-canonical-per-decision-LLM-self-consistency-cross-check + per-canonical-per-decision-LLM-chain-of-thought-extraction.

What does per-decision FBC-feedback + per-cross-skill-handoff do?

Per-portfolio per-decision per-canonical-FBC-feedback-loop runs per-portfolio per-canonical-FBC-per-decision-realized-outcome-vs-projected-outcome-feedback + per-canonical-FBC-per-decision-reviewer-approval-vs-LLM-approval-feedback + per-canonical-FBC-per-decision-policy-decision-vs-LLM-decision-feedback + per-canonical-FBC-per-decision-routing-decision-pattern-learning + per-canonical-FBC-per-decision-policy-rule-drift-detection + per-canonical-FBC-per-decision-prompt-template-drift-detection + per-canonical-FBC-per-decision-model-drift-detection + per-canonical-FBC-per-decision-completion-drift-detection + per-canonical-FBC-per-decision-cost-vs-budget-feedback + per-canonical-FBC-per-decision-latency-vs-SLO-feedback + per-canonical-FBC-per-decision-reviewer-SLA-vs-realized-feedback + per-canonical-FBC-per-decision-policy-circuit-breaker-recalibration + per-canonical-FBC-per-decision-emerging-failure-mode-detection. Per-canonical-per-decision-cross-skill-handoff runs per-portfolio per-canonical-per-decision-handoff-to-audit-trail-software (parent commercial pillar at /audit-trail-software) + per-canonical-per-decision-handoff-to-ai-routing-decision-audit-trail (sibling commercial at /ai-routing-decision-audit-trail) + per-canonical-per-decision-handoff-to-versioned-history-regulatory-defense (sibling build-pillar at /how-to-build-versioned-history-regulatory-defense-for-multi-location-operators) + per-canonical-per-decision-handoff-to-integration-health-monitoring (sibling build-pillar at /how-to-build-marketing-stack-integration-health-for-multi-vendor-campaign-operations) + per-canonical-per-decision-handoff-to-attribution-rollup (sibling build-pillar at /how-to-build-per-location-cross-channel-attribution-rollup-for-franchise-reporting) + per-canonical-per-decision-handoff-to-per-location-metric-ingestion (sibling build-pillar at /how-to-build-per-location-metric-ingestion-across-10-15-sources-per-location) + per-canonical-per-decision-handoff-to-pos-receipt-integration + per-canonical-per-decision-handoff-to-versioned-product-history (sibling build-pillar at /how-to-build-versioned-product-history-for-recall-traceability-across-multi-vendor-supply-chains) + per-canonical-per-decision-handoff-to-brand-voice-management + per-canonical-per-decision-handoff-to-forbidden-phrase-library + per-canonical-per-decision-handoff-to-customer-data-graph.

What does per-decision audit-trail + per-governance-decision-router-agent-canonical-bundle do?

Per-portfolio per-decision per-canonical-audit-trail runs per-portfolio per-canonical-per-decision-canonical-audit-record (per-decision-ID + per-banner-pointer + per-location-pointer + per-route-pointer + per-model-pointer + per-prompt-template-version + per-system-message-snapshot + per-tool-call-trace + per-RAG-retrieval-snapshot + per-vector-store-pointer + per-embedding-model-pointer + per-temperature + per-top-p + per-max-tokens + per-seed + per-completion-snapshot + per-tokens-in-out + per-cost + per-latency + per-policy-engine-trace + per-policy-version-pointer + per-policy-decision-log + per-prompt-completion-fingerprint + per-PII-PHI-redaction-snapshot + per-reviewer-role + per-reviewer-sign-off-state + per-reviewer-comment + per-reviewer-SLA + per-multi-LLM-pre-publish-record + per-FBC-feedback per-canonical-audit-record) + per-canonical-EU-AI-Act-Article-13-transparency + per-canonical-EU-AI-Act-Article-14-human-oversight + per-canonical-EU-AI-Act-Article-15-accuracy-robustness-cybersecurity + per-canonical-NIST-AI-RMF-attestation + per-canonical-NIST-SP-800-218A-secure-AI-development + per-canonical-NIST-SP-800-53-control-objective-attestation + per-canonical-SOC-2-Type-II + per-canonical-ISO-27001 + per-canonical-ISO-42001-AI-management-system + per-canonical-HIPAA-PHI + per-canonical-PCI-DSS + per-canonical-FedRAMP + per-canonical-FTC-substantiation-doctrine + per-canonical-FTC-MARS-multi-location-substantiation + per-canonical-FTC-AI-disclosure + per-canonical-FDA-AI-ML-software-as-medical-device + per-canonical-FINRA-2210-broker-dealer + per-canonical-CFPB-UDAAP + per-canonical-CCPA-CPRA-DSAR + per-canonical-GDPR-DPIA + per-canonical-LGPD-ANPD + per-canonical-DPDP-India + per-canonical-PIPEDA-OPC + per-canonical-CASL-CRTC + per-canonical-FDD-Item-12-Item-17-Item-19 + per-canonical-immutable-WORM-storage + per-canonical-7-year-tax-audit-IRS-retention. Per-governance-decision-router-agent-canonical-bundle integrates the routing-audit-trail skill with sibling skills on the same governance-decision-router agent: per-canonical-routing-audit-trail (this skill) + per-canonical-audit-trail-software (parent commercial pillar) + per-canonical-ai-routing-decision-audit-trail (sibling commercial pillar) + per-canonical-policy-as-code-OPA-Rego-AWS-Cedar-Casbin-Cerbos-Oso (sibling) + per-canonical-prompt-template-version-management (sibling) + per-canonical-model-version-management (sibling) + per-canonical-reviewer-workflow (sibling) + per-canonical-PII-PHI-redaction (sibling) + per-canonical-vector-store-version-management (sibling) + per-canonical-RAG-retrieval-trace (sibling). Per-canonical-end-to-end-SLA runs per-canonical-per-decision-routing-capture-to-policy-trace-to-prompt-completion-snapshot-to-reviewer-sign-off-to-pre-publish-to-present-to-FBC-SLA canonical-SLA.

Engage the governance-decision-router agent

Per-portfolio per-banner per-location per-decision per-canonical-routing-capture-spec + per-canonical-policy-engine-trace-spec + per-canonical-prompt-completion-snapshot-spec + per-canonical-reviewer-sign-off-spec + per-canonical-multi-LLM-pre-publish-check + per-canonical-FBC-feedback-loop + per-canonical-audit-trail + per-portfolio audit-trail shipped as the orchestration layer above your existing per-LLM-observability-vendor + per-AI-guardrail-vendor + per-PII-PHI-redaction-vendor + per-AI-governance-platform-vendor primitive.

Book a scope call Browse the 32-agent catalog