minimal-repro

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

• Third-party content exposure detected (high risk: 0.75). The skill ingests “TypeSpec source (pasted or GitHub URL)” and “Error description” from the operating user, but at runtime it will fetch and read outsider-authored free text from a GitHub URL (public web content) into the LLM context via the provided source/error text.