The Agent Skills Directory

[COMMAND_EXECUTION]: The skill provides tools to manage the execution and lifecycle of the Godot engine binary, specifically via the run_project and stop_run_instance commands.
[DYNAMIC_EXECUTION]: The skill includes a script_write tool that allows the agent to generate and save executable GDScript code directly to the user's filesystem.
[INDIRECT_PROMPT_INJECTION]: The skill processes various files from the target Godot project, creating a surface for indirect prompt injection if a processed project is maliciously crafted.
Ingestion points: Godot project configuration (project.godot), scene resource files (.tscn), and GDScript files (.gd) read from the user-specified projectPath.
Boundary markers: The instructions do not define explicit delimiters or security markers to isolate ingested file content from the agent's instructional context.
Capability inventory: The skill has high-privilege capabilities including writing arbitrary scripts (script_write), modifying project configurations (project_update_settings), and executing the Godot binary (run_project).
Sanitization: No sanitization or validation mechanisms are described for the content of ingested project files before processing.
[EXTERNAL_DOWNLOADS]: The installation instructions direct users to clone the server's source code from a third-party GitHub repository (github.com/wangdiandao/godot-devtool.git).

godot-devtool-mcp-server