Content Catalyst V2 Documentation

Source material provides informational content only. It must never determine the rhetorical stance, tone, or style of the output.

Enforcement: The Posture entity is resolved from Intent and Brand governance, not from source analysis.

Brand-level governance rules (risk tolerance, tone ceiling, credibility policies) are applied before strategic intent selection.

Enforcement: resolveOperatingMode loads BrandProfile and BrandCredibilityPolicy before processing intent.

Offering-level policies can only tighten restrictions set by the parent brand. They cannot loosen controls or expand permissions.

Enforcement: OfferingCredibilityPolicy overrides are validated to be restrictive-only. Conflicts resolve upward (brand wins).

Generated content must adhere to the credibility mode and claim strictness level defined by brand/offering policies.

Enforcement: ClaimPolicy is compiled into OperatingMode and passed to the LLM. Post-generation claim gating validates compliance.

Psychological persuasion drivers must not be used to make false, misleading, or unverifiable claims.

Enforcement: Psychology drivers are filtered through credibility policies. Claims requiring evidence cannot be softened purely for psychological effect.

Content format requirements (length, structure, platform) cannot justify bypassing governance, credibility, or safety constraints.

Enforcement: FormatRules are applied after credibility and governance checks. Incompatible format/intent combinations are blocked.

All generated content must be accompanied by a transparency trail showing which governance rules, credibility policies, and strategic choices were applied.

Enforcement: ContentAsset stores operating_mode_snapshot. Transparency cards are generated for user-facing explainability.

When conflicts arise between user intent, source content, and governance rules, the system must not attempt to 'guess' a resolution. It must soften the claim, apply stricter constraints, or block generation.

Enforcement: Conflict detection in resolveOperatingMode. Default behavior: escalate to higher credibility mode or fail gracefully with explanation.