The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill references multiple external GitHub repositories as recommended MCP servers for 3D modeling and AI image generation (e.g., ahujasid/blender-mcp, AIDC-AI/Pixelle-MCP, joenorton/comfyui-mcp-server). These represent code dependencies from third-party sources not included in the trusted provider list.
[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and evaluate data from external design tools and AI generation pipelines, creating an attack surface for indirect prompt injection.
Ingestion points: Pulls UI design context from Figma files, reviews AI-generated concept art and textures, and interacts with 3D scene data from Blender.
Boundary markers: The instructions do not define boundary markers or explicit guidelines for the agent to ignore potentially malicious instructions embedded within the metadata, comments, or textual components of these external assets.
Capability inventory: The skill is granted Write and Edit tools, allowing it to modify local project files (such as the Art Bible or Style Guide) based on the input received from these untrusted external sources.
Sanitization: There is no mention of sanitization or validation protocols for content retrieved from external design APIs or generated by AI models.

game-art-director