flux-image
Pass
Audited by Gen Agent Trust Hub on Mar 17, 2026
Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill allows the agent to execute
infshcommands using theBashtool to interact with the inference.sh API. This is the intended functionality for running remote image generation models. - [EXTERNAL_DOWNLOADS]: The skill documentation references the installation of the
infshCLI and additional skills from the vendor's official repository (inference-sh/skills). These are recognized as trusted vendor resources. - [PROMPT_INJECTION]: The skill has a surface for indirect prompt injection through the processing of user-provided image prompts.
- Ingestion points: Untrusted user input is passed into the
promptfield of the JSON input for theinfsh app runcommand as seen inSKILL.mdexamples. - Boundary markers: Absent; there are no delimiters or instructions to the model to ignore potential commands embedded within the user's image prompt.
- Capability inventory: The skill has the capability to execute commands on the inference.sh platform via the
Bashtool. - Sanitization: The provided documentation does not include input validation or sanitization for the prompt strings.
Audit Metadata