blender-hardsurface

CRITICAL E006: Malicious code pattern detected in skill scripts.

• Malicious code pattern detected (high risk: 1.00). The repository exposes a local JSON socket API (localhost:9876) with an "execute_code" command that will run arbitrary Python inside Blender, providing a clear remote-code-execution / backdoor capability if the socket is reachable or abused locally.

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

• Third-party content exposure detected (medium risk: 0.30). The runtime workflow listens on a local JSON socket and the outsider can supply arbitrary free text/code via the execute_code command (e.g., send("execute_code", {"code": ...})), which is directly ingested by BlenderMCP and executed.