The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructs the agent to execute several shell commands using the renku CLI tool to perform model discovery and retrieve operation schemas. Specifically, it uses renku generation models list and renku generation schema show. These commands rely on arguments provided by the skill's internal files and potentially external discovery outputs.
[INDIRECT_PROMPT_INJECTION]: The skill's design involves reading and acting upon external data sources which could be manipulated to influence agent behavior. This includes provider-supplied documentation, model guides located in references/model-guides/, and dynamic 'personal notes' fetched via CLI commands.
Ingestion points: Data is ingested from the references/supported-routes.json file, model-specific guides in the references/model-guides/ directory, and external provider documentation URLs.
Boundary markers: The instructions specify the use of $file markers to explicitly identify and delimit local media file references within the generated native requests.
Capability inventory: The skill possesses the capability to execute shell commands (via the renku CLI) and perform file system writes within the project's tmp/ directory structure.
Sanitization: There is no explicit mention of sanitizing or validating the content retrieved from external documentation or personal guides before it is used to influence the construction of generation commands.

wavespeed-media-provider