anyrouter

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

• Third-party content exposure detected (high risk: 0.85). At runtime the skill instructs the agent to “fetch the relevant .md URL” from https://anyrouter.dev/docs*.md / llms*.txt (public web content) and read it as markdown, which is outsider-authored text injected into the LLM context.

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

• Potentially malicious external URL detected (high risk: 0.90). The skill explicitly requires fetching live markdown docs at runtime (e.g., https://anyrouter.dev/docs.md and related .md URLs) and instructs the agent to ground its responses in that fetched content, so the external URL(s) directly control prompts/instructions and are a required runtime dependency.