The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: Fetches installation instructions and documentation from the official GitHub repository for inference.sh.
[COMMAND_EXECUTION]: Executes the belt command-line tool to perform video generation and account authentication.
[INDIRECT_PROMPT_INJECTION]: The skill processes user-defined prompts for video generation, which could theoretically contain instructions intended to influence model behavior. 1. Ingestion points: prompt parameter in JSON inputs passed to the belt CLI tool in SKILL.md. 2. Boundary markers: No delimiters are used to wrap user input within the prompts. 3. Capability inventory: The skill's environment is restricted to the belt CLI as defined in the allowed-tools frontmatter. 4. Sanitization: There is no indication of input sanitization or filtering in the documentation.

google-veo