cx-coding-agents

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

• Third-party content exposure detected (high risk: 0.75). Outsider-authored free text can enter the LLM context via the “Claude Code (session text)” runtime path: cx logs ... source ai_sessions_claude ... choose ... $d.parsed.text/flat_text ... -o agents, where $d.* contains turn-by-turn conversation messages (user/assistant content) that are not authored by the operating user.