adcp-creative
Pass
Audited by Gen Agent Trust Hub on Oct 9, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data through the natural language
messagefield and asset definitions within creative manifests, which could be used to embed malicious instructions. - Ingestion points: Data enters the agent's context through user-provided instructions in
build_creativeand external asset URLs/content increative_manifestobjects. - Boundary markers: The skill does not define specific delimiters or instructions to prevent the agent from following commands that might be embedded in the creative data.
- Capability inventory: The skill utilizes tools to fetch, transform, and render content based on these inputs, which includes making network requests to specified agent URLs.
- Sanitization: No explicit sanitization or validation of the input messages or external asset contents is provided in the instructions.
- [EXTERNAL_DOWNLOADS]: The skill references and fetches creative format specifications and delivery data from the vendor's official domain at
creative.adcontextprotocol.org.
Audit Metadata