The Agent Skills Directory

[PERSISTENCE]: The skill instructs the agent to modify the user's shell profile (~/.bashrc) by appending an export command to the PATH variable. This ensures that the locally installed Node.js binaries remain accessible in all future shell sessions.
[COMMAND_EXECUTION]: The instructions require the agent to execute a wide variety of shell commands, including curl, tar, npm, and the qhkit CLI utility, to set up the environment and perform video generation tasks.
[EXTERNAL_DOWNLOADS]: The skill directs the agent to download the Node.js runtime from nodejs.org and the qhkit package from the npm registry. While these are well-known services, downloading and extracting binaries at runtime is a significant capability.
[INDIRECT_PROMPT_INJECTION]: The skill processes user-provided text prompts and local images to generate video content, creating a surface where malicious instructions could potentially be embedded in the input data.
Ingestion points: Text prompts and image file paths provided in the JSON arguments for video generation commands in SKILL.md.
Boundary markers: None present to distinguish between instructions and data.
Capability inventory: Shell command execution, network uploads via the CLI tool, and file system access.
Sanitization: No explicit sanitization or validation of the prompt content is performed before processing.

qinghu-ecom-video