policy-govern

Advancedworkflow

Mission

Audit → enforce → monitor → remediate. Zero tolerance for undetected compliance violations.

When to Use

Use when auditing AI workflows for policy compliance, enforcing data guardrails, validating model governance, hardening against prompt injection, designing regulated workflows, monitoring compliance drift, or remediating security and governance issues.

Triggers: “compliance check”, “safety audit”, “policy validation”, “data guardrails”, “prompt injection hardening”, “regulated workflow”, “governance review”, “model version policy”

Skills Invoked

Two-pass pattern (GPT-5.4 first-pass, Claude Sonnet 4.6 final judgment):

• gov-prompt-injection-hardening — injection vector detection and mitigation
• gov-policy-validation — validate against defined policy spec
• gov-data-guardrails — PII detection, scrubbing, purpose limitation
• gov-model-governance — approved model audit
• gov-model-compatibility — model roster compatibility check
• gov-regulated-workflow-design — HIPAA/GDPR/SOC 2 workflow design
• gov-workflow-compliance — end-to-end compliance check

Chain-To

• code-review — address code-level issues found in governance audit
• fault-resilience — add structural fault tolerance recommended by governance
• docs-generate — generate compliance documentation

Strict Mode

When ENABLE_GOVERNANCE_STRICT=true:

• Human-in-the-loop checkpoint inserted before results are returned
• All gov-* skills require approval before proceeding

Example

{
  "request": "Audit our customer data processing pipeline for GDPR compliance"
}

Output: Compliance gap table, data flow map with PII exposure points, remediation priority list, and a compliance certificate once all gaps are resolved.