blender-mcp-setup
Pass
Audited by Gen Agent Trust Hub on Aug 22, 2026
Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill uses the
mcp__blender__execute_blender_codetool to interact with the local Blender process. This is the core purpose of the skill and is implemented with safety guidelines such as disabling automatic script execution for untrusted scenes as noted inSKILL.md.- [PROMPT_INJECTION]: The skill processes scene data (object names, file paths) from Blender, which could potentially contain malicious instructions embedded in scene metadata. - Ingestion points:
mcp__blender__get_scene_info,mcp__blender__get_object_info, andmcp__blender__get_viewport_screenshotinSKILL.md. - Boundary markers: Not present; instructions rely on external safety skills for untrusted content.
- Capability inventory: Python execution authority within Blender via
mcp__blender__execute_blender_codeinSKILL.md. - Sanitization: Absent; instructions rely on non-destructive probes and manual review of code chunks.
Audit Metadata