defuddle

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

• Third-party content exposure detected (high risk: 0.95). Outsider-authored free text is fetched from a runtime-supplied public URL (arbitrary web page content) and converted to readable markdown by npx -y defuddle parse --markdown "$url", which is then fed into the agent/LLM context via the tool’s stdout.

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

• Potentially malicious external URL detected (high risk: 1.00). The script runs npx -y defuddle ... at runtime which fetches and executes the remote npm package (e.g. from https://registry.npmjs.org/defuddle), so required external code is downloaded and executed during runtime.