The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructs the agent to perform various system-level operations, including software installation, version control tasks with git, and management of Unity Editor processes.
[DYNAMIC_EXECUTION]: The skill provides the script eval command, which allows the agent to execute arbitrary C# code directly within a running Unity Editor. This is a powerful feature intended for automating scene and asset modifications.
[CREDENTIALS_UNSAFE]: The unity auth token command can expose active access tokens to the standard output. Additionally, several commands accept sensitive information like API tokens and keystore passwords via command-line arguments or environment variables.
[DATA_EXFILTRATION]: The unity bug command is designed to collect and transmit diagnostic data, logs, and project files to the vendor for troubleshooting.
[PERSISTENCE]: The skill facilitates the installation of plugins and session-start hooks in AI agents (e.g., Claude Code, Codex) to maintain tool integration across sessions.
[INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data from Unity Editor logs and git history. Malicious instructions could be embedded in this data to attempt to manipulate the agent's behavior during command execution.
[EXTERNAL_DOWNLOADS]: The skill provides instructions to download the CLI and its components from the vendor's official domains and trusted repositories.

unity-cli