event-prospecting

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

• Third-party content exposure detected (medium risk: 0.30). The workflow ingests arbitrary web pages via browse cloud fetch and browse open / browse get markdown across multiple steps (recon, page extraction, and web searches for target companies and speakers), exposing the agent to outsider-authored free text without prior selection.

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

• Potentially malicious external URL detected (medium risk: 0.30). The compatibility and instructions require installing the browse CLI from npm (npm install -g browse), which acts as an external tool dependency used throughout the skill's execution.