The Agent Skills Directory

[SAFE]: The skill provides tools and instructions for Godot 4.7 gameplay development without any detected malicious patterns.
[INDIRECT_PROMPT_INJECTION]: The skill processes external data formats such as .dialogue files and .tres save resources. However, it implements proactive security measures:
Ingestion points: Processes game save resources and dialogue files in scripts/gd_gameplay.py and save_io.gd.
Boundary markers: The skill provides guidance on data-only design and uses specific Godot loading modes (CACHE_MODE_IGNORE) to maintain data integrity.
Capability inventory: Includes file system access and shell command execution (via the peer skill scenario-godot-expert) for benchmarks and tests.
Sanitization: Implements scan_resource_text and load_resource_checked which use regular expressions to detect and block embedded scripts (GDScript, CSharpScript) in text-based Godot resources, mitigating potential code execution via malicious save files.
[COMMAND_EXECUTION]: The skill executes Godot scripts and benchmarks using a standard runner from its associated environment. This behavior is consistent with the skill's primary purpose for gameplay systems development.
[EXTERNAL_DOWNLOADS]: The documentation references established open-source projects (Dialogue Manager, LimboAI, GUT) and official Godot documentation. No automated downloads or execution of untrusted remote code were detected.

scenario-godot-gameplay