shadcn

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

• Third-party content exposure detected (high risk: 0.90). The skill (SKILL.md and cli.md) explicitly instructs the agent to fetch and read component docs/examples and upstream component files from external URLs (e.g., ui.shadcn.com docs links, raw GitHub/example URLs, npx shadcn@latest add https://...), and mcp.md documents registry URLs and MCP tools that return full file contents from public/community registries—untrusted third-party content the agent must parse and that can change subsequent CLI actions.

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

• Potentially malicious external URL detected (high risk: 0.90). The skill explicitly instructs fetching component docs/examples at runtime (e.g., raw content URLs such as https://raw.githubusercontent.com/...) and those fetched raw files are used to drive the agent's instructions and project changes, so runtime external content can directly control prompts/code.