Skillget

codex-delegate

agent
Security Audit
Pass
Health Pass
  • License — License: MIT
  • Description — Repository has a description
  • Active repo — Last push 0 days ago
  • Community trust — 50 GitHub stars
Code Pass
  • Code scan — Scanned 2 files during light audit, no dangerous patterns found
Permissions Pass
  • Permissions — No dangerous permissions requested
Purpose
This tool acts as a bridge to route heavy coding tasks—such as multi-file refactoring and boilerplate generation—from Claude to the OpenAI Codex CLI. It allows Claude to write a task file, execute Codex synchronously via a wrapper script, and review the resulting code changes before final acceptance.

Security Assessment
Overall Risk: Low. The tool acts primarily as a task supervisor and script runner. Its wrapper scripts inherently execute shell commands to launch the external Codex CLI, which is its intended purpose. It does not request dangerous permissions, and the automated code scan found no hardcoded secrets, malicious patterns, or hidden network requests. Since it routes code instructions to a third-party CLI (OpenAI), standard precautions regarding the data you pass into the external service still apply.

Quality Assessment
The project is in excellent shape and demonstrates active maintenance, with its most recent push occurring just today. It uses the permissive MIT license and has garnered 50 GitHub stars, indicating a healthy level of community trust for a specialized agent. The repository is cleanly organized, provides clear documentation in multiple languages, and includes regression tests for its wrapper scripts.

Verdict
Safe to use.
SUMMARY

Claude Code skill for supervising Codex CLI on implementation-heavy coding tasks, multi-file refactors, and wrapper-based execution flows.

README.md

Codex Delegate

繁體中文

codex-delegate is a Claude-oriented skill for routing implementation-heavy work to Codex CLI while keeping planning, review, and acceptance in Claude.

Positioning

This skill is for tasks that are expensive in tokens but cheap in judgment:

  • multi-file implementation
  • mechanical refactors
  • boilerplate generation
  • test scaffolding
  • large batch edits

It is not meant for architecture, root-cause debugging, security review, or ambiguous product decisions.

What Changed In This Version

  • clearer routing boundary between Claude, Codex, and Gemini
  • explicit supervisor acceptance gate
  • machine-readable wrapper output via <log>.result.json
  • regression tests for bash and PowerShell wrappers

Core Pattern

  1. Claude writes a task file describing scope and constraints.
  2. Claude launches Codex synchronously through the wrapper.
  3. The wrapper emits sentinel files plus result.json.
  4. Claude reviews the diff and runs verification before accepting the result.

Wrapper success is not final acceptance. Claude still owns the judgment.

Repository Layout

codex-delegate/
├── SKILL.md
├── README.md
├── README_zh-TW.md
├── scripts/
│   ├── run_codex.sh
│   └── run_codex.ps1
├── tests/
│   └── test_wrappers.py
└── references/

Testing

python -m pytest -q

Current wrapper tests cover:

  • success-path result.json generation
  • PowerShell wrapper contract behavior

Installation

Codex CLI must be available in your environment:

npm install -g @openai/codex
codex --version

License

MIT

Reviews (0)

No results found