mage-marketplace-cards
Pass
Audited by Gen Agent Trust Hub on Nov 11, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted user data, including product names, features, and claims, and interpolates this content into prompts for image generation models (Mango 4, GPT Image 2.5 Flare).
- Ingestion points: User-supplied product facts and listing data collected during the workflow defined in SKILL.md.
- Boundary markers: The skill uses structured prompt templates and a 'Product lock' block in
references/card-recipes.mdto delimit user content. - Capability inventory: The skill uses Mage generation tools (
generate,get_model,create_upload) via the MCP protocol. - Sanitization: There is no evidence of input sanitization or filtering to prevent malicious instructions from being passed to the downstream image models.
- [EXTERNAL_DOWNLOADS]: The skill references and requires the use of the official Mage infrastructure (
mcp.mage.space) for the MCP connector and user accounts. These resources are provided by the skill's author.
Audit Metadata