triage

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

• Third-party content exposure detected (high risk: 0.75). The required workflow only instructs the agent to “Read the full request” (e.g., message/Linear/Slack/bug report) and then classify it; those inputs are typically outsider-authored free text delivered at runtime via the user’s linked ticket/chat, which the LLM would ingest into its context.