The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses the Bash tool to interact with the belt CLI. The allowed-tools configuration limits bash usage exclusively to the belt command, which is a recommended security practice for restricting the agent's execution environment.
[EXTERNAL_DOWNLOADS]: The skill references external resources for installation and extended functionality, such as https://raw.githubusercontent.com/inference-sh/skills/refs/heads/main/cli-install.md. These links are official sources for the platform described and do not involve immediate or hidden code execution.
[INDIRECT_PROMPT_INJECTION]: The skill accepts user-defined prompts for image generation which are passed to an external API via the CLI. 1. Ingestion points: The prompt and input parameters in various belt app run examples in SKILL.md. 2. Boundary markers: The prompt content is encapsulated within JSON structures when passed to the shell command. 3. Capability inventory: The skill is authorized to execute the belt binary through the restricted Bash tool. 4. Sanitization: The skill relies on the underlying CLI and API for input validation, as no explicit filtering is defined in the instruction markdown.

ai-image-generation