read-docs

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

• Third-party content exposure detected (high risk: 0.90). The skill fetches and parses content directly from external Fumadocs sites (via /llms.txt discovery, raw .mdx pages such as https://example.com/docs/... .mdx, and the /api/search JSON endpoint) and uses that untrusted public documentation as the basis for answers, which could enable indirect prompt injection.

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

• Potentially malicious external URL detected (high risk: 0.90). The skill explicitly fetches external documentation at runtime (e.g., https://example.com/docs/installation.mdx, the site root /llms.txt, and /api/search?query=...) and directly ingests that content into the agent's prompt/context to drive responses, so these endpoints are runtime dependencies that can control prompts.