codex-control-plane-mcp

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

• Third-party content exposure detected (high risk: 0.70). The skill’s runtime ingests OUTSIDER-authored free text via the MCP tool inputs like codex_submit_task/steer_turn/codex_approve_plan where the operating user may pass arbitrary prompts or feedback originating from external parties; that text is then used by the control plane to drive the Codex app-server/LLM context.