deepline-pre-research

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

• Third-party content exposure detected (medium risk: 0.30). The required workflow monitors and consumes outsider-authored social feeds, public discussions, and community posts (such as Reddit threads/comments and X/Twitter posts) without the agent first selecting a specific item.

MEDIUM W012: Unverifiable external dependency detected (runtime URL that controls agent).

• Potentially malicious external URL detected (high risk: 0.70). An operational dependency exists to code.deepline.com to install and register packages, but it utilizes an uncontrolled non-official domain and custom path during runtime setup, classifying it as a weak operational dependency.