ad-creative
Pass
Audited by Gen Agent Trust Hub on Jun 14, 2026
Risk Level: SAFEPROMPT_INJECTIONCOMMAND_EXECUTIONEXTERNAL_DOWNLOADS
Full Analysis
- [PROMPT_INJECTION]: The skill analyzes external performance data (CSV, paste, or API output) to iterate on ad copy, which presents a surface for indirect prompt injection.
- Ingestion points: Performance data enters the agent context through the "Iterating from Performance Data" workflow in
SKILL.md. - Boundary markers: The skill does not define explicit delimiters or instructions to disregard potential commands embedded within the performance data.
- Capability inventory: The skill interacts with external generative AI APIs and local advertising CLI tools (
node tools/clis/google-ads.js). - Sanitization: No explicit sanitization or validation logic is provided for the ingested data before it is processed by the model.
- [COMMAND_EXECUTION]: The skill documents the use of local CLI tools and shell commands (
curl,npx) to interact with advertising platforms and generative AI services. These commands are standard for automated marketing workflows and are used as intended for the skill's primary purpose. - [EXTERNAL_DOWNLOADS]: Includes references to official generative AI APIs (Google Gemini, ElevenLabs, OpenAI) and provides installation instructions for well-known open-source creative tools like Remotion and Voicebox from their respective official GitHub repositories and domains.
Audit Metadata