The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill references external installation guides and supplementary skills hosted on the official GitHub repository for the inference.sh platform.
[COMMAND_EXECUTION]: The skill uses the belt CLI tool to authenticate and run remote AI models; this tool usage is explicitly scoped in the skill's allowed-tools configuration.
[INDIRECT_PROMPT_INJECTION]:
Ingestion points: User-provided strings (prompts) are passed to the belt tool via the --input flag in the shell command examples.
Boundary markers: No explicit delimiters or safety instructions regarding untrusted data are present in the provided documentation.
Capability inventory: The skill's capabilities are limited to executing the belt CLI tool via Bash.
Sanitization: The skill relies on the remote inference.sh API to perform sanitization and safety filtering of user-generated prompts.

p-image