codex-delegate

1# Codex Delegate
2 
3You are the **orchestrator**. This skill lets you hand a bounded coding task to a separate
4**implementer** — the OpenAI Codex CLI — then review what it produced and land it yourself. You write
5the brief and own the judgment; Codex does the typing in its own sandbox; you verify and commit.
6 
7Nothing here is specific to one orchestrating agent. The loop needs only the ability to run a shell
8command and read a file, so it works the same whether you are Claude Code, OpenCode with a selected
9model, or any comparable agent. (It is designed for and run on Claude Code; treat other orchestrators
10as designed-for, not yet proven.)
11 
12## When NOT to use this
13 
14- The task is small enough to just do inline — delegation overhead is not worth it.
15- The `codex` CLI is not installed or not authenticated (run `codex login`).
16- You want to write the code yourself, or you only need a review (use Codex's own `review` command).
17 
18## Prerequisites (check once)
19 
201. `codex --version` succeeds. If not, install (`npm i -g @openai/codex`) and `codex login`.
212. **Confirm which `codex` is on PATH.** Multiple installs are common (e.g. a current npm/nvm copy and
22   a stale Homebrew one). `command -v codex` shows the active one and `codex --version` its version —
23   an old binary predates flags this skill relies on (`codex exec --json`, `-o`, `exec resume`). The
24   relay also records the version it ran into `result.json`, so a stale binary is visible after the fact.
25…