The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill exhibits a surface for indirect prompt injection because it reads untrusted data from external files and possesses tools that can modify the filesystem.
Ingestion points: The shader_read skill is designed to ingest the full source code of .shader files into the agent's context.
Boundary markers: The instructions do not specify the use of delimiters or specific warnings to the agent to treat content from shader_read as data rather than instructions.
Capability inventory: The skill provides powerful capabilities including file creation (shader_create, shader_create_urp), file deletion (shader_delete), and global state modification (shader_set_global_keyword).
Sanitization: There is no documented mechanism for sanitizing or validating the HLSL/ShaderLab content read from files before it is processed by the agent.

unity-shader