daub-ui

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

• Third-party content exposure detected (medium risk: 0.30). In the DAUB Playground/MCP workflow described by PIPELINE.md, the agent ingests arbitrary outsider-authored userPrompt text (and attached webContexts text) via buildMessages(...) before streaming generation and subsequent interactivity analysis.

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

• Potentially malicious external URL detected (high risk: 0.90). The skill explicitly instructs agents to connect to a remote MCP endpoint (https://daub.dev/api/mcp) and to fetch LLM-facing docs/data (https://daub.dev/llms.txt and https://daub.dev/components.json) at runtime, which are external endpoints that can directly influence agent prompts/tool outputs.