What problem does it solve? Enterprises generating AI imagery at scale risk off-brand or non-compliant assets reaching production. This Skill gates every generation through a written brand policy, so prompts are pre-validated, outputs are post-checked, violations escalate to human approvers, and every decision is logged for audit. ## Core Features & Use Cases - Policy-as-code gating: Authors a versioned brand.md (palette, typography, allowed/denied props, regulated-category rules) whose git commit SHA becomes the policy ID in every audit record. - Three-checkpoint enforcement: Pre-flight prompt linting via gen-ai validate, brand-context injection into gen-ai generate and gen-ai batch run, and post-generation vision audits with gemini-3-pro-image. - Escalation and audit trail: Routes violations to an approver queue, logs every decision as approved/flagged/rejected, and exports daily JSONL to a SIEM or compliance archive. - Use Case: A pharmaceutical marketing team runs a 1,000-image campaign batch; the governor rejects prompts containing prohibited claims before credits are spent, vision-audits hero assets, and ships the full decision log to Splunk for compliance review. ## Quick Start Ask the agent to validate a campaign prompt against your brand.md policy file and then run a gated batch generation with violations escalated to your brand approver.
