brand-voice

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

• Third-party content exposure detected (high risk: 0.90). The skill's Mode 3 "Auto-Scrape" explicitly fetches and scrapes public third‑party content (user-provided URL plus homepage, about page, blog posts, LinkedIn/Twitter/X) via Exa MCP and then directly analyzes that untrusted user-generated/public content to build a voice profile that will influence downstream actions, so it clearly ingests and acts on open web content.

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

• Potentially malicious external URL detected (high risk: 0.90). The skill's Auto-Scrape mode fetches a user-provided URL at runtime (e.g., https://sahilbloom.com) and injects the scraped content into the model's analysis flow to generate prompts/samples, so external content directly controls the agent's output.