The Agent Skills Directory

[COMMAND_EXECUTION]: The skill leverages the allowed-tools metadata to restrict shell access exclusively to the belt command. This scopes the agent's capabilities to only the intended CLI tool, reducing the attack surface for arbitrary command execution.
[EXTERNAL_DOWNLOADS]: The skill recommends the installation of the belt-sh/cli package and references configuration files from the inference-sh GitHub repository. These are legitimate dependencies for the platform integration described in the skill's purpose.
[SAFE]: No patterns of prompt injection, data exfiltration, or obfuscation were identified. The use of external image URLs for image-to-image tasks is a standard feature of the underlying model and does not pose an inherent security risk in this context.

flux-image