How I Got Claude to Consult Codex (and When I Use It)
Up until recently, getting Claude to collaborate with other agents was…functional, but clunky.
My regular workflow looked like this: I’d ask Claude to write a file listing open questions, what we're doing, trying or considering, then hand that file off to another agent and request a response. It worked well for pulling in extra perspectives, but it was slow and inefficient.
I’ve tried a few different approaches including various MCP servers and landed on a skill that allows Claude to directly consult Codex when "we’re" stuck or when I want a second opinion.
The Shift: Claude as Lead, Codex as Counsel
Instead of bouncing artifacts between agents, Claude stays in charge of the conversation. When progress stalls—or when I want parallel work—I just ask Claude to consult /codex.
How to Install the Codex Skill
- Go to your home directory
- Press
Cmd + Shift + .(to show hidden files) - Find the
.claudefolder and inside that create a folder calledskills - Clone or copy this repo inside it (make sure the skill.md and references are present)
Once Claude sees the skill, you’re good. You can test it by typing /skills and should see confirmation.

You can now:
- Use
/codexdirectly while prompting with Claude - Or ask Claude to delegate something to
/codexon your behalf
Somewhere I stumbled on someone suggesting using it in a ralph-loop or a CLAUDE.md and they shared this snippet:
CODEX DELEGATION (OPTIONAL):
- Use /codex <task> or codex exec --full-auto <task> for parallel work.
- Good candidates for Codex delegation:
- Writing tests for code you just wrote
- Implementing a component while you work on another
- Code review of completed work items
- Do NOT delegate to Codex when:
- The task depends on something you're currently building
- Multiple agents would edit the same file
- The task requires your current conversation context
- When delegating, run Codex in background and continue your work.
- Check Codex output before marking work item complete.Working Example
