The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill references the installation of the belt CLI and related functionality from external repositories, specifically pointing to npx skills add belt-sh/cli and documentation hosted on raw.githubusercontent.com/inference-sh/.
[COMMAND_EXECUTION]: The skill uses the Bash tool to execute belt commands, which are scoped to the belt binary as defined in the allowed-tools frontmatter. These commands handle authentication (belt login) and interface with image generation models (belt app run).
[INDIRECT_PROMPT_INJECTION]: The skill provides a surface for processing user-defined prompts and external image URLs that are passed to the Gemini models.
Ingestion points: User-provided text prompts and image URLs are accepted via the --input argument in various belt app run examples in SKILL.md.
Boundary markers: No explicit delimiters or instructions are used to separate user input from model instructions within the skill's documentation.
Capability inventory: The skill is authorized to execute belt CLI commands via the Bash tool to communicate with remote inference services.
Sanitization: Input prompts and image links are passed directly to the CLI tool without intermediate validation or sanitization within the skill's instructions.

nano-banana