star-content-agent

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

• Third-party content exposure detected (medium risk: 0.30). The agent reads user bookmarks from platforms like Xiaohongshu and Douyin via a browser, ingesting outsider-authored content (titles, text, comments) without user pre-selection.

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

• Potentially malicious external URL detected (high risk: 0.70). The skill instructs the agent to download and install third-party packages and extensions from arbitrary or uncontrolled GitHub repositories and URLs (e.g., npm install -g @jackwener/opencli, GitHub releases, and Chrome Web Store) at runtime during setup, constituting a weak/unknown/suspicious operational dependency.