codex-insights

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

• Third-party content exposure detected (medium risk: 0.30). The skill’s required runtime workflow reads outsider-authored free text from locally stored Codex session files (e.g., user messages) by scanning $CODEX_HOME/~/.codex for session_index.jsonl, sessions/**/*.jsonl, and archived_sessions/**/*.jsonl, and extracting message content / first_user_message into the report.