roadmap: structured policy extensions for content standards

[bokelley]

Context

The Creative Working Group's Brand Manifesto & Standards Library framework identifies several gaps in how brand rules are expressed in AdCP. Most of the framework's proposed objects already exist (brand.json, content standards, campaign governance), but four areas lack structured schema support — rules are expressible only as free-text policy today.

This roadmap covers extending content standards (and potentially brand.json) to close those gaps. Each item should become its own issue when ready for implementation.

Items

1. Structured data and consent policy

Priority: High — the largest real gap in the protocol.

Clean room rules, inference prohibitions, and first-party data permissions have no structured representation. Agents cannot programmatically enforce data boundaries — they can only read free-text policy and hope to interpret it correctly.

Fields to consider:

• first_party_data_uses — what first-party data can be used, where, and for what purposes
• inference_prohibitions — explicit prohibitions on certain types of profiling or inference
• clean_room_rules — matching rules and restrictions for clean room environments
• consent_requirements — consent signals required before data activation

This could be a new object on content standards or a standalone protocol object referenced by brand. Needs design work before implementation.

2. Risk tiers as structured fields

Priority: Medium — extends existing governance with per-topic granularity.

Campaign governance has audit/advisory/enforce modes per layer, but no per-topic risk classification. Rather than a separate risk_profile object, add structured fields to content standards:

• risk_level — enum (low, medium, high, critical)
• review_required — boolean, triggers human review before agent action
• review_reason — why review is required (for audit trail)

This lets a brand say "politics → high risk, require review" as a queryable field rather than buried in policy text.

3. Effective dates on content standards

Priority: Medium — important for policy rollouts and sunsets.

Content standards have versioning but no activation/expiration dates.

• effective_from — ISO 8601 datetime, when this policy becomes active
• effective_until — ISO 8601 datetime, when this policy expires (optional)

Agents should only apply policies where the current time falls within the effective window.

4. Adjacent categories as a discovery signal

Priority: Low — needs scoping before implementation.

Adjacent categories help agents avoid unexpected brand adjacencies — a children's cereal brand next to alcohol content, for example. The Creative WG flagged this as a "strong near-term candidate" and references a separate proposal. Should become its own issue once the WG scopes the design.

Design principles

• Extend existing objects (content standards, brand.json) rather than introducing new top-level manifest objects. Avoid creating competing sources of truth.
• Keep fields machine-readable and queryable — the whole point is moving beyond free-text.
• Each item ships independently — don't bundle these into one mega-PR.

Relationship to existing work

• Brand Manifesto & Standards Library framework (Creative WG)
• Content Standards protocol (shipped)
• Campaign governance with layer hierarchy (shipped, Governance cross-refs and protocol simplifications #1772, feat!: sync_governance task, remove governance from sync/list #1644)
• Visual tokens on brand.json (feat: visual_guidelines — border radius, elevation, spacing, extended color roles #1871, visual_guidelines: add border radius, elevation, spacing, and color granularity #1864)