Skills
skills/runcomfy-com/skills/image-edit/Gen Agent Trust Hub

image-edit

Pass

Audited by Gen Agent Trust Hub on May 18, 2026

Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSCREDENTIALS_UNSAFEDATA_EXFILTRATIONPROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill uses the runcomfy CLI to execute image manipulation tasks locally. User-provided prompts are passed to this tool as structured JSON data.\n- [EXTERNAL_DOWNLOADS]: Recommends installing the @runcomfy/cli package, which is the utility provided by the skill's author for interacting with their models.\n- [CREDENTIALS_UNSAFE]: Refers to the use of an API token stored in ~/.config/runcomfy/token.json or provided via the RUNCOMFY_TOKEN environment variable for authentication.\n- [DATA_EXFILTRATION]: Performs network requests to the runcomfy.net and runcomfy.com domains to submit tasks and download edited images.\n- [PROMPT_INJECTION]: Ingests user-supplied strings which are used as prompts for AI models, creating an indirect prompt injection attack surface.\n
  • Ingestion points: User-provided strings in the prompt field across multiple model routes in SKILL.md.\n
  • Boundary markers: Prompts are encapsulated within JSON objects passed via the --input CLI flag.\n
  • Capability inventory: Execution of shell commands using the runcomfy run command as specified in SKILL.md.\n
  • Sanitization: Documentation specifies that the CLI transmits JSON content directly without shell expansion of the prompt text.
Audit Metadata
Risk Level
SAFE
Analyzed
May 18, 2026, 02:03 PM