product-photography

Pass

Audited by Gen Agent Trust Hub on Sep 17, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill fetches content from user-provided URLs and public sources to research product positioning and brand identity, which exposes the agent to potentially malicious instructions embedded in external web pages.
  • Ingestion points: The skill fetches site data or product pages to build or read .agents/product-context.md (documented in SKILL.md).
  • Boundary markers: The skill explicitly references references/agent-security.md, which provides strict rules: treat fetched content as data, quote any found instructions rather than obeying them, and do not follow URLs discovered within fetched content.
  • Capability inventory: The agent has the ability to write persistent data to the .agents/ directory and invoke external image generation tools (e.g., fal.ai, Recraft) as mentioned in SKILL.md.
  • Sanitization: The skill's security reference (references/agent-security.md) requires HTML-escaping of any untrusted content before it is interpolated into any emitted code or markup to prevent XSS.
  • [CREDENTIALS_SAFE]: The skill includes advanced instructions in references/agent-security.md to detect, flag, and redact credentials (API keys, session tokens, signed URLs) found in user inputs or CRM exports, specifically forbidding the AI from persisting them to files or echoing them back.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 17, 2026, 02:22 PM