Starts and manages an OpenAI Codex CLI session so an agent can perform deep-code reasoning, review functions, and analyze code. The skill decides whether to start a new session or resume an existing one, immediately returns a task id, and instructs the user how to fetch results.
Use this skill when a user asks for code analysis, bug-finding, architectural review, or other deep reasoning tasks that benefit from a dedicated Codex session (e.g., "Find bugs in this function", "Explain the auth flow"). Triggered by commands like /codex.
start_codex_session or resume_codex_session depending on whether the conversation continues, returns a task id immediately, and tells the user how to check results with /codex-result <task_id>. It explicitly avoids blocking/waiting for the Codex run to finish.Likely compatible with agents that can call external MCP tools and manage background tasks (CLI-driven agents, Copilot-like agents, Claude Code integrations).
This skill has not been reviewed by our automated audit pipeline yet.
