The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses the allowed-tools field to restrict shell execution to the infsh command. This follow the principle of least privilege by limiting the agent's capability to a specific CLI tool.
[EXTERNAL_DOWNLOADS]: The documentation references installing the inference-sh/skills@agent-tools package via npx skills add. This is an external dependency within the vendor's own ecosystem for managing agent capabilities.
[PROMPT_INJECTION]: As the skill passes user-provided prompts directly to the infsh app run command via JSON input, there is a surface for indirect prompt injection where malicious input could attempt to influence the video generation model's output. This is a common risk for generative AI integrations.
[DATA_EXFILTRATION]: The skill mentions infsh login for handling authentication. This is a standard procedure for the service and does not involve hardcoded credentials or unauthorized access to sensitive local files.

google-veo