local-codex

An agent skill by daymade, from daymade/claude-code-skills. Tags: automation, cli, code, local-workflow, multi-agent.

What it does

Launch and manage OpenAI Codex CLI (local agent) as a non-interactive coding sub-agent. Use when the user wants to delegate coding tasks to Codex, run code reviews, generate or refactor code, or use Codex GPT-5.5 agent capabilities through local CLI. Triggers on phrases like 'codex', 'run codex', 'codex exec', 'code review with codex', 'delegate to codex', 'use codex for coding', or any request to invoke the local Codex CLI agent. Uses ChatGPT Pro OAuth (flat-rate, no API charges) via ~/.codex/auth.json. Never uses API keys.

Install

With the skills CLI, which installs into Claude Code, Codex, Cursor and other agents:

npx skills add daymade/claude-code-skills --skill local-codex

Or copy the skill folder into Claude Code's skills directory by hand (~/.claude/skills for every project, or .claude/skills inside one):

git clone --depth 1 https://github.com/daymade/claude-code-skills
cp -r claude-code-skills/daymade-codex/local-codex ~/.claude/skills/local-codex

Safety box score

Grade D, elevated risk: 46 of 100 risk points, rated 2026-10-08. The grade rates what the skill and its scripts can reach on the machine of whoever installs it, across eight categories from shell execution to secrets access. How the score works.

This skill is a local wrapper around the Codex CLI: it launches an external `codex` binary to run non-interactive coding tasks and reviews. The main risks are remote data leaving the machine, reuse of OAuth tokens in `~/.codex/auth.json`, and the fact that the wrapper can invoke an opaque external executable with whatever sandbox is chosen.

Category levels, from 0 (none) to 3 (high)
Category	Level	Why
Code & shell execution	3 (high)	`scripts/codex_wrapper.py` directly runs `[codex_path, "exec", ...]` and `review` via `subprocess.run`, so it can trigger arbitrary agent-managed code execution.
Network access	2 (moderate)	The skill is built around `ChatGPT Pro OAuth` and `codex login` / `codex exec`, so prompts and workspace context are sent to the external OpenAI/Codex service even though there is no raw URL fetch in the repo.
Filesystem reach	2 (moderate)	It reads and uses home-directory state such as `~/.codex/auth.json` and `~/.codex/logs_*.sqlite`, and accepts a caller-chosen workdir with `-C`, reaching outside the project tree.
Secrets & credentials	2 (moderate)	`references/oauth-guide.md` says `Tokens are cached in ~/.codex/auth.json` and that the desktop app and CLI share that cache, so the skill touches a refresh-token store.
Privilege & persistence	0 (none)	There is no `sudo`, service install, global package install, or shell/OS settings mutation in the shipped files.
Instruction hijack surface	0 (none)	The files contain CLI usage guidance but no instructions to ignore safety rules, follow fetched instructions, or hide information from the user.
Irreversible actions	1 (low)	The wrapper runs Codex non-interactively and can let it edit a workdir, but it does not itself push, deploy, send messages, pay, or book anything.
Transparency	2 (moderate)	The wrapper chooses an external executable from `CODEX_BIN`, `which codex`, or `/Applications/ChatGPT.app/.../codex`, so the actual behavior depends on code not included in this repo.

Before installing

Check which binary will run: `CODEX_BIN`, then `which codex`, then the ChatGPT.app bundle.
Read `references/oauth-guide.md` and confirm you're comfortable with `~/.codex/auth.json` being the shared auth cache.
Test first in a throwaway workdir with `sandbox=read-only` before allowing any write access or `danger-full-access`.

Rated 2026-10-08 by an automated review of 3 files from the skill's repository.

Source

Repository: daymade/claude-code-skills (all skills from this repository)
Path: daymade-codex/local-codex/SKILL.md
Branch: main
Updated: 2026-09-20

bambu-labs — Dry-run, upload, and cautiously initiate local Bambu Lab print jobs from validated plain `.gcode`, using Bambu LAN FTPS/MQTT handoffs.
dxf — Generate, regenerate, and validate 2D DXF drawings from Python build123d sources.
E2E Testing Nori Skillsets Subcommands — Use when you need to interactively test a nori-skillsets CLI subcommand end-to-end via tmux, with full filesystem isolation
filesystem-context — This skill should be used when agent work needs file-backed context: durable scratchpads, tool-output offloading, just-in-time discovery.
gcode — Generate, inspect, dry-run, and statically validate plain FDM `.gcode` from 3D mesh files by orchestrating real slicer CLIs. Use when Codex needs to slice.
agents-build — Use to extend an existing agent project with memory, app integration, VPC, multi-agent, migration, model, browser, code interpreter, payments.