The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest data from external sources, such as reference images for detail planning and documentation like 'uv-mapping-dayz.md', to inform the 3D modeling process. This creates a surface for indirect prompt injection where instructions hidden in external data could influence the Python code generated by the agent. Ingestion points: Reference images and external project documentation (e.g., 'dayz-model-pipeline'). Boundary markers: The skill lacks explicit markers or instructions for the agent to disregard instructions within processed data. Capability inventory: Access to 'mcp__blender__execute_blender_code' allows for arbitrary Python execution in the host environment. Sanitization: No sanitization or validation of data retrieved from external sources is provided before its use in geometry calculations.
[DYNAMIC_EXECUTION]: The instructions explicitly mandate the reading and execution of the 'references/blender_helpers.py' file at the start of every session. While the script performs standard geometry operations, this bulk execution of a local script file increases the complexity of auditing individual operations performed via the MCP.
[COMMAND_EXECUTION]: The skill heavily utilizes high-privilege tools to execute Python code within the Blender application to manage geometry, mesh integrity, and texture baking.

blender-assembly