The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes local CLI commands and Node.js scripts (e.g., blender-toolkit, node .blender-toolkit/bt.js) to interact with the Blender process. These commands are necessary for the skill's primary function of 3D automation.
[DATA_EXPOSURE]: The skill accesses a configuration file at ~/.claude/plugins/marketplaces/dev-gom-plugins/blender-config.json and maintains logs in the .blender-toolkit/ directory. This is standard practice for managing plugin state and troubleshooting.
[EXTERNAL_DOWNLOADS]: The instructions guide the user to download Blender from its official website (blender.org) and fetch animation files from Adobe's Mixamo service. These are well-known, trusted services required for the 3D workflow.
[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted external files and environment data.
Ingestion points: FBX animation files ingested via the retarget command's --file argument, and object/material data retrieved from the Blender WebSocket server.
Boundary markers: None explicitly defined for parsing data returned from the local WebSocket server or the animation files.
Capability inventory: The skill can execute local shell commands and scripts, and modify files within the Blender environment.
Sanitization: There is no explicit mention of sanitization or validation for the data extracted from FBX files or Blender object lists before they are processed by the agent.

blender-toolkit