The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill instructs the user or agent to install an unverified third-party package, unity-mcp-cli, globally via NPM (npm install -g unity-mcp-cli) or to run it via npx. This package is not from a trusted organization or the recognized vendor, and no specific version or integrity checksum is provided to ensure safety.
[COMMAND_EXECUTION]: The skill uses a subprocess to execute the unity-mcp-cli tool. It passes user-supplied search filters and folder paths directly into the command line arguments. While it suggests using stdin or input files to mitigate injection risks, the underlying execution mechanism remains shell-based.
[INDIRECT_PROMPT_INJECTION]: The skill possesses a vulnerability surface where malicious instructions could be embedded in the data it processes.
Ingestion points: Data is ingested from the Unity project asset database (filenames, labels, and types) in response to the filter parameter.
Boundary markers: No delimiters or instructions to ignore embedded content are provided in the prompt templates.
Capability inventory: The skill has the capability to execute shell commands via the unity-mcp-cli tool.
Sanitization: There is no evidence of sanitization, validation, or escaping of the asset metadata returned from the database before it is presented to the agent.

assets-find