design-md

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

• Third-party content exposure detected (high risk: 0.90). The SKILL.md workflow explicitly instructs the agent to download and parse HTML and screenshots from htmlCode.downloadUrl and screenshot.downloadUrl (via web_fetch / read_url_content) returned by the Stitch MCP Server—these are user-provided/untrusted project assets that the agent must read and whose content is used to drive analysis and generation of the DESIGN.md, enabling indirect instruction injection.