The Agent Skills Directory

[REMOTE_CODE_EXECUTION]: The skill provides an installation command that pipes a remote script from the vendor's domain to the shell (curl -fsSL https://cli.inference.sh | sh). This is a documented and expected pattern for installing the inference.sh CLI utility required by the skill.
[COMMAND_EXECUTION]: The skill uses the Bash tool to execute infsh commands. These commands are used to authenticate with the service and run video generation models with user-provided prompts.
[EXTERNAL_DOWNLOADS]: The skill fetches configuration and binaries from the vendor's official distribution servers at cli.inference.sh and dist.inference.sh.
[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection as it interpolates untrusted user data into command-line arguments for the infsh tool. \n
Ingestion points: User-provided prompt strings used in the --input flag of the infsh app run command in SKILL.md.\n
Boundary markers: The prompt is enclosed within a JSON string inside a single-quoted bash argument, providing basic structural separation.\n
Capability inventory: Access to the Bash tool with permissions to execute infsh commands and related subcommands.\n
Sanitization: No explicit sanitization or filtering of the prompt content is performed within the skill code itself.

google-veo