llms-txt-support

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

• Third-party content exposure detected (high risk: 0.80). The skill explicitly fetches and ingests llms.txt files from arbitrary public websites (e.g., https://example.com/llms.txt, llms-full.txt, llms-small.txt and .well-known locations) and uses that content as documentation to read/interpret, exposing the agent to untrusted third-party content that could contain embedded instructions.

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

• Potentially malicious external URL detected (high risk: 1.00). The skill explicitly fetches llms.txt files from target sites at runtime (e.g., https:///llms.txt as shown by the curl examples) and injects that external document into the agent's context, meaning remote content can directly control prompts/instructions.