OpenAI Codex Sub Agents

Use OpenAI Codex CLI for coding tasks. Triggers: codex, code review, fix CI, refactor code, implement feature, coding agent, gpt-5-codex. Enables Clawdbot to delegate coding work to Codex CLI as a subagent or direct tool.

Install this only if you want Clawdbot or another agent to delegate coding work to Codex CLI. Use it in trusted repositories, set `--cd` to a specific project, prefer read-only or approval-based modes for unfamiliar code, avoid `danger-full-access` and `--yolo` unless you fully accept the risk, and review Codex auth/session storage and Clawdbot token sync before enabling it.

Type: OpenClaw Skill Name: codex-sub-agents Version: 1.0.0 The skill bundle introduces the `codex CLI` tool, which grants extensive local system access, including arbitrary command execution and file system modifications. The `SKILL.md` and `clawdbot-integration.md` files explicitly instruct the OpenClaw agent to use `codex exec --full-auto`, enabling workspace writes and command execution without explicit approval. Furthermore, the documentation (`SKILL.md`, `cli-reference.md`) details highly privileged flags like `--sandbox danger-full-access` (full machine access) and `--yolo` (no approvals or sandbox), making the agent aware of these dangerous capabilities. While there's no clear evidence of intentional malicious behavior like data exfiltration to external endpoints, the broad system access and instruction to use high-privilege commands without explicit user approval classify this as suspicious due to significant inherent risks.