anthropic

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

• Third-party content exposure detected (high risk: 0.70). This skill’s required workflow is an Anthropic SDK integration that sends user-provided message/content strings (e.g., request.message or messageContent) as the messages payload to Claude, so any outsider-authored free text supplied by the operating user at runtime would be ingested into the LLM context.