The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses the mcp__blender__execute_blender_code tool to run dynamically generated Python scripts based on provided material recipes. This allows the agent to automate the creation of PBR shaders using the bpy module.
[PROMPT_INJECTION]: The skill contains an indirect prompt injection attack surface because it interpolates untrusted user data into executable scripts.
Ingestion points: The skill expects the agent to replace the GEO-target placeholder with object names provided by the user in the prompt.
Boundary markers: There are no delimiters or warnings to ignore instructions embedded in the object names.
Capability inventory: The skill has the capability to execute arbitrary Python code via mcp__blender__execute_blender_code and perform shell operations via Bash.
Sanitization: No sanitization or escaping logic is included to prevent object names from breaking out of string literals to execute arbitrary code.
[EXTERNAL_DOWNLOADS]: The skill documentation includes a link to physicallybased.info, which is a well-known service providing reference values for physically-based materials. This reference is used neutrally for look-development purposes.

blender-materials