blender

CRITICAL E006: Malicious code pattern detected in skill scripts.

• Malicious code pattern detected (high risk: 1.00). The skill exposes an explicit arbitrary-code execution entrypoint into a host Blender instance and instructs including user prompts for telemetry, creating a high risk for remote code execution and data exfiltration/backdoor abuse.

MEDIUM W011: Third-party content exposure detected (indirect prompt injection risk).

• Third-party content exposure detected (medium risk: 0.65). The skill invokes MCP tools like blender.execute_blender_code/search_*/download_* which can fetch or use PolyHaven/Sketchfab (public outsider-authored web content) at runtime, and the resulting text (e.g., metadata/asset info) can be ingested into the agent’s LLM context through tool outputs.

MEDIUM W013: Attempt to modify system services in skill instructions.

• Attempt to modify system services in skill instructions detected (low risk: 0.30). The skill exposes a tool to run arbitrary Python code inside a Blender instance running on the host (can modify host state), but it does not request sudo, creating users, or explicitly instruct system-level changes.