friction-log

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

• Third-party content exposure detected (medium risk: 0.30). The workflow requires the agent to read any URLs provided in the user's prompt (GitHub repos, documentation, reference files), as well as ingest out-of-band harness messages (such as chat-thread replies or queued Slack messages), allowing outsider-authored text to be consumed at runtime.