generating-product-photos
Pass
Audited by Gen Agent Trust Hub on Aug 22, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill is composed entirely of markdown documentation and reference guides. It does not contain any executable scripts (Python, JavaScript, shell), nor does it attempt to install external dependencies.- [INDIRECT_PROMPT_INJECTION]: The skill defines a workflow for processing user-supplied product images and URLs, creating a surface for potential indirect prompt injection.
- Ingestion points: The workflow (Step 1) ingests external data in the form of product images or URLs which are analyzed using vision tools (
analyzing-products,image_analysis). - Boundary markers: The skill incorporates fixed prompt blocks across all modes that enforce product identity preservation ("The attached image is the exact product. Do not modify, regenerate or redraw it") and delineate scene construction from product description.
- Capability inventory: The agent utilizes the
image_generatetool to produce new visual outputs based on analyzed product data. - Sanitization: The skill mitigates risks by instructing the agent to focus on physical properties, materials, and lighting characteristics extracted via specialized analysis tools, rather than verbatim instruction following from the source material.
Audit Metadata