agent-browser

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

• Potentially malicious external URL detected (high risk: 0.80). The skill runs runtime fetches like "agent-browser skills get core" (and similar "agent-browser skills get electron/slack/…") which download upstream skill docs at runtime and are then parsed/extracted to drive agent behavior, so these external fetches can directly control prompts/instructions.