ai-image-generation

Pass

Audited by Gen Agent Trust Hub on May 19, 2026

Risk Level: SAFEPROMPT_INJECTION
Full Analysis
  • [PROMPT_INJECTION]: The skill possesses an attack surface for indirect prompt injection where malicious instructions embedded in external data could steer the model's output.
  • Ingestion points: Untrusted data enters the workflow through image_urls, images, and the enable_web_search feature described in SKILL.md.
  • Boundary markers: The skill instructs the agent to encapsulate inputs within JSON structures and provides explicit warnings to the operator regarding the untrusted nature of external reference assets.
  • Capability inventory: The skill is scoped to execute the runcomfy CLI using the Bash tool as defined in the allowed-tools frontmatter.
  • Sanitization: The documentation indicates the CLI transmits JSON directly over HTTPS without shell expansion, though it acknowledges that the underlying AI models can still be influenced by content within reference images or search results.
Audit Metadata
Risk Level
SAFE
Analyzed
May 19, 2026, 06:45 PM