codex-monitor

You can monitor Codex CLI workflows by using a runtime source that applies persistent runtimeSessionId management across turns, calling MCP tools like monitor_runtime_session_ensure and monitor_assistant_response to orchestrate end-to-end tasks and maintain traceability.

When the monitor-server is unavailable, Agent Tracer monitoring continues the workflow tasks uninterrupted and generates a final gap report at the end to ensure degraded connectivity remains observable.

To manage a persistent runtime session ID across Codex turns, use a codex-skill runtimeSource that calls monitor_runtime_session_ensure at the start and monitor_runtime_session_end upon completion to coordinate the workflow.

Yes, workflow discovery and coordination for similar tasks is supported during Codex-driven task monitoring, ensuring full traceability and observability throughout the orchestrated workflow lifecycle.

Yes, you need MCP tools to coordinate end-to-end Codex monitoring, as the workflow relies on calling monitor_runtime_session_ensure, monitor_assistant_response, and monitor_runtime_session_end to orchestrate tasks and report gaps.

Codex workflow monitoring requires gap reporting to maintain observability when the monitor-server experiences degraded connectivity, ensuring any unmonitored portions of the runtime session are documented at the end.