relight

Pass

Audited by Gen Agent Trust Hub on Jul 13, 2026

Risk Level: SAFEPROMPT_INJECTION
Full Analysis
  • [SAFE]: The skill utilizes the @runcomfy/cli package and runcomfy.com endpoints, which are confirmed vendor resources associated with the skill author prime-skills.
  • [PROMPT_INJECTION]: The skill processes external image URLs and text prompts, which constitutes an indirect prompt injection surface.
  • Ingestion points: User-provided image URLs and lighting descriptions are ingested as input for the CLI tool (SKILL.md).
  • Boundary markers: Input data is encapsulated within a JSON string passed via the --input argument to prevent shell expansion.
  • Capability inventory: Command execution is scoped to the runcomfy CLI tool through the Bash interface.
  • Sanitization: The documentation explicitly states that the CLI does not shell-expand prompt content, minimizing the risk of command injection.
Audit Metadata
Risk Level
SAFE
Analyzed
Jul 13, 2026, 04:10 AM
Security Audit — agent-trust-hub — relight