skills/inf-sh/skills/flux-image/Gen Agent Trust Hub

flux-image

Pass

Audited by Gen Agent Trust Hub on Apr 4, 2026

Risk Level: SAFEREMOTE_CODE_EXECUTIONCOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [REMOTE_CODE_EXECUTION]: The documentation includes a command to install the platform's CLI utility by fetching a script from cli.inference.sh and piping it to the shell. This is the official installation method for the vendor's tooling.\n- [COMMAND_EXECUTION]: The skill performs image generation using the infsh command-line tool. The agent's execution environment is restricted to this specific tool via platform configuration, preventing arbitrary command execution.\n- [PROMPT_INJECTION]: The skill processes user-provided prompts which are passed as arguments to the CLI tool, creating a surface for indirect instructions.\n
  • Ingestion points: User prompts are ingested through the --input argument in SKILL.md examples.\n
  • Boundary markers: Input is structured within JSON strings to separate user data from command flags.\n
  • Capability inventory: The skill's capabilities are limited to executing the infsh CLI tool.\n
  • Sanitization: No explicit sanitization of the prompt strings is performed before they are passed to the service API.
Audit Metadata
Risk Level
SAFE
Analyzed
Apr 4, 2026, 09:10 PM
Security Audit — agent-trust-hub — flux-image