gpt-image-edit

Pass

Audited by Gen Agent Trust Hub on Aug 13, 2026

Risk Level: SAFE
Full Analysis
  • [COMMAND_EXECUTION]: The skill uses the runcomfy CLI to interact with the RunComfy Model API. The documentation specifies that the CLI transmits JSON bodies directly without shell-expanding user prompts, mitigating injection risks.\n- [CREDENTIALS_UNSAFE]: The skill manages API tokens using environment variables or restricted-permission configuration files (~/.config/runcomfy/token.json). No hardcoded secrets were found.\n- [EXTERNAL_DOWNLOADS]: The skill requires the @runcomfy/cli package, which is a vendor-owned resource necessary for the skill's functionality.\n- [INDIRECT_PROMPT_INJECTION]: The skill processes external image URLs. The documentation includes a dedicated security section warning about image-based prompt injection and provides preservation prompting patterns to mitigate this risk. \n
  • Ingestion points: images array in JSON input.\n
  • Boundary markers: Explicit instructions to 'Keep' specific attributes and quote text.\n
  • Capability inventory: Remote image editing via Model API.\n
  • Sanitization: Prompt JSON is not shell-expanded.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 13, 2026, 10:57 AM
Security Audit — agent-trust-hub — gpt-image-edit