ad-creative-generation
Pass
Audited by Gen Agent Trust Hub on Oct 29, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONNO_CODE
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill extracts branding data, personality traits, and UI descriptions from user-provided URLs using the
firecrawl_branding_extracttool. This external content is directly interpolated into prompts for generating ad copy and image generation requests, creating a potential vector for indirect prompt injection attacks where a malicious website could influence the agent's output. - Ingestion points: The
firecrawl_branding_extracttool inSKILL.mdandreferences/brand-extraction.mdfetches content from external URLs. - Boundary markers: The instructions do not specify any boundary markers (e.g., XML tags or delimiters) or explicit 'ignore instructions' warnings when processing the extracted brand personality or screenshot descriptions.
- Capability inventory: The skill has the capability to generate text (ad copy) and images (via
images_generate) based on the untrusted data. - Sanitization: There is no mention of sanitization, filtering, or validation of the data retrieved from the website before it is used in downstream tasks.
- [NO_CODE]: The skill consists entirely of Markdown instructions and reference files. No executable scripts (Python, JavaScript, etc.) are included in the skill package.
Audit Metadata