image-generation-enhanced

Pass

Audited by Gen Agent Trust Hub on Mar 12, 2026

Risk Level: SAFEPROMPT_INJECTION
Full Analysis
  • [PROMPT_INJECTION]: The skill instructions establish an attack surface for indirect prompt injection by directing the agent to ingest and process untrusted user-provided prompt strings and external files.
  • Ingestion points: User-provided prompts, reference images, and external files (e.g., '@prompt.md') referenced in SKILL.md.
  • Boundary markers: No protective delimiters or warnings to ignore embedded instructions are included in the workflow.
  • Capability inventory: The skill leverages the 'minibanana' CLI tool to execute image generation tasks using these untrusted inputs.
  • Sanitization: There are no explicit instructions for validating, escaping, or sanitizing the input data before it is passed to the generation tool.
Audit Metadata
Risk Level
SAFE
Analyzed
Mar 12, 2026, 02:09 PM