blender

CRITICAL E004: Prompt injection detected in skill instructions.

• Potential prompt injection detected (medium risk: 0.30). The skill sets up Blender Lab's add-on from official, hash-pinned sources with disclosed consent, but it also opens Blender without asking and forces online access for the session, which weakens a user protection beyond what the task needs, so it is risky without showing malicious intent.

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

• Attempt to modify system services in skill instructions detected (medium risk: 0.40). The skill does not request sudo, system-file edits, account creation, or service or SSH changes, and its installs stay in user-owned folders with consent. It still leads the agent to install software and an add-on, change Blender's saved preferences, launch Blender with a flag that grants online access, and run an unauthenticated Python bridge on port 9876, so I flag it at moderate severity.