The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill provides instructions to install the belt CLI and associated skill packages from the official inference-sh GitHub organization and NPM registry. These downloads are necessary for the skill's primary function of interacting with the inference.sh platform.
[COMMAND_EXECUTION]: The skill instructions utilize the belt command-line interface to perform inference tasks. The execution environment is limited to the belt tool as defined in the allowed-tools frontmatter configuration.
[INDIRECT_PROMPT_INJECTION]: The skill processes user-provided text prompts and external image URLs to generate or modify images. This creates a potential surface for indirect prompt injection if the processed content contains instructions for the model.
Ingestion points: User input passed via the --input argument in shell commands within SKILL.md.
Boundary markers: The use of structured JSON for input parameters helps delineate data from commands.
Capability inventory: The skill uses shell execution to invoke the belt CLI for remote API calls.
Sanitization: Sanitization is handled by the backend inference service and the CLI tool rather than the skill instructions themselves.

p-image