blender

CRITICAL E006: Malicious code pattern detected in skill scripts.

• Malicious code pattern detected (high risk: 1.00). This code intentionally exposes an unauthenticated localhost HTTP server that compiles/executes arbitrary Python in the live Blender process and can be installed to auto-start — enabling local RCE, easy persistence, and multiple straightforward data-exfiltration channels.

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

• Third-party content exposure detected (medium risk: 0.30). SKILL.md describes a required runtime that executes arbitrary Python “POSTed to 127.0.0.1:8736” via the Blender Bridge HTTP server, meaning outsider-authored free text (the posted {code}/{path} content) is ingested and exec’d at runtime.