blender-workflow-routing
Pass
Audited by Gen Agent Trust Hub on Nov 14, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONDYNAMIC_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to process untrusted or externally supplied scene content, which represents an indirect prompt injection attack surface.
- Ingestion points: Untrusted scene content is imported into Blender (SKILL.md, Step 1).
- Boundary markers: Not explicitly defined within this routing skill, though it delegates to a safety-specific skill.
- Capability inventory: Python execution (
blender-python-execution), file system access (save/export), and network operations via MCP. - Sanitization: The skill mandates the use of
blender-content-safety-and-privacyfor incoming data andblender-sanitized-support-bundlefor outgoing logs and reports. - [EXTERNAL_DOWNLOADS]: The instructions describe a workflow that includes installing optional upstream community production skills and recipes (e.g.,
blender-community-skill-bootstrap,text-to-blender). These represent external dependencies from potentially untrusted community sources. - [COMMAND_EXECUTION]: The skill orchestrates the setup of MCP (Model Context Protocol) servers and the installation of the OVRTX add-on, involving local environment modifications.
- [DYNAMIC_EXECUTION]: The skill facilitates the execution of Python code within the Blender environment (
blender-python-execution), which is standard for Blender automation but requires careful handling of input.
Audit Metadata