The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses shell commands like mkdir, cp, and export to set up a new Godot project structure. It also invokes the godot binary in headless mode to verify project integrity and run initial smoke tests.
[INDIRECT_PROMPT_INJECTION]: The skill accepts user-provided project slugs and absolute paths which are interpolated into shell commands for scaffolding.
Ingestion points: User-defined PROJECT_DIR and slug variables in the provided shell snippets.
Boundary markers: The instructions recommend inspecting the destination first, but no explicit sanitization is performed on the variables.
Capability inventory: The skill performs file system writes (mkdir, cp) and subprocess execution (godot).
Sanitization: The skill relies on the agent's shell execution environment for parameter safety.
[DATA_EXPOSURE]: The skill implements security best practices by redirecting Godot's configuration and data directories to a project-local .godot-xdg folder using XDG_DATA_HOME, XDG_CONFIG_HOME, and XDG_CACHE_HOME environment variables. This prevents the agent from reading or polluting the user's global Godot environment.

scaffolding-godot-mini-games