remotion-video

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

• Third-party content exposure detected (high risk: 0.75). Outsider-authored free text can be ingested when the skill uses Playwright MCP to scrape Google Business Profile reviews (e.g., testimonial.md Step 1 extracts “3 best… reviews” including actual review text) and then uses that scraped text in the subsequent LLM-driven scene/script/code generation.

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

• Potentially malicious external URL detected (high risk: 0.80). The skill explicitly requires using Playwright MCP at runtime to visit and scrape user-provided product or profile URLs (e.g., "https://your-product-url.com" and "https://maps.google.com/some-business"), and the scraped page content is used directly to drive the agent's prompts, assets, and build steps—constituting a runtime external dependency that controls the agent's behavior.