Spec Driven Development
MCP Server that facilitates spec-driven development workflows, not just Vibe Coding.
Description from the maker (original English)
MCP server wrapping Codex CLI for stdio-based tool calls.
These three are never merged into one score: opinions, measurements and popularity are different things.
We have not written a plain-language summary for this one yet. Below are the facts we read from the registry.
Pick the AI app you use, then copy the settings.
You need
Which app do you use?
Pick your app above. We remember it, so next time the button says “Install in …”.
Open Claude Desktop → Settings → Developer → Edit Config, and paste this into claude_desktop_config.json
{
"mcpServers": {
"codex-go": {
"command": "npx",
"args": [
"-y",
"@zenfun510/codex-mcp-go"
]
}
}
}Run this in your terminal
claude mcp add codex-go -- npx -y @zenfun510/codex-mcp-goPaste into .cursor/mcp.json in your project, or ~/.cursor/mcp.json
{
"mcpServers": {
"codex-go": {
"command": "npx",
"args": [
"-y",
"@zenfun510/codex-mcp-go"
]
}
}
}Your browser asks to open the app, then the app shows the settings and asks you to confirm. Key values are left empty for you to fill in.
Paste into .vscode/mcp.json in your project
{
"servers": {
"codex-go": {
"type": "stdio",
"command": "npx",
"args": [
"-y",
"@zenfun510/codex-mcp-go"
]
}
}
}Your browser asks to open the app, then the app shows the settings and asks you to confirm. Key values are left empty for you to fill in.
This one runs on your computer, and ChatGPT can only connect to servers on the internet. Pick another app.
Your choice stays on this device in a cookie; no account needed.
These settings are generated from the maker’s registry entry. If the README says otherwise, follow the README.
Runs on your computer; the tool list appears in your AI app after you install it.
We check information anyone can see; we have not read the code. Checks that do not apply are left out of the count. · Checked Nov 7, 2026
The registry name was verified with the GitHub account w31r4, which owns the repo
We have not read this repo from GitHub yet
We have not read this repo from GitHub yet
This one runs on your computer; there is no internet endpoint
Runs on your computer and uses local settings instead
Runs on your computer; we can only read tool lists from internet endpoints
Compared with every active server in this category in the registry. Stars count only when the publisher owns the repo.
It passed 1 of 1 applicable basic checks. We have not read the code. Before adding a key or token, see what access the server asks for and use a key with only the permissions you need.
The registry does not list a required key, though some servers ask you to sign in on first use.
It runs on your computer. Your AI app starts it when needed, and data does not pass through the maker’s servers (unless it calls an online service itself).
The GitHub repo last changed –, and the latest version (0.2.1) reached the registry 9 months ago.
Apps that support MCP, such as Claude Desktop, Claude Code, Cursor and VS Code. You also need Node.js on your computer.
In the same category: Spec Driven Development, Remote Desktop Commander, MemberVault MCP by usefulapi
Servers in the same category from other makers, sorted by stars.
Reading a card:★ GitHub starsiThree pillars: reviews · checks · usageiUpdated = latest version published to the registryi
MCP Server that facilitates spec-driven development workflows, not just Vibe Coding.
Remote Desktop Commander: use files, terminal and processes on your own computers from any AI app.
List courses; add users to courses, remove course access or delete users in MemberVault.
Copied from the maker’s repo as written. PopNx MCP has not reviewed or endorsed it. HTML, images and scripts are removed. Links lead off this site.
Codex CLI 的 MCP 协议封装实现
⭐ 如果觉得好用,请给个 Star 吧!您的支持是我们更新的动力~ ⭐
English | 简体中文
codex-mcp-go 是一个基于 Go 语言实现的 MCP (Model Context Protocol) 服务器。它封装了 OpenAI 的 Codex CLI,使其能够作为 MCP 工具被 KiloCode、Roo Code、Cline 等各种 "Vibe Coding" AI 客户端调用。
这个项目的初衷很简单:让强者更强,让专才专用。
在我自己的工作流中(在 KiloCode 里使用 Gemini 3.0 Pro),我发现像 Gemini 这样的先进模型拥有强大的规划能力和想象力,但在修复自己生成的复杂代码时偶尔会“卡壳”。而 Codex,这位老练的“代码老师傅”,虽然在宏大叙事上稍逊一筹,但在具体的代码实现、Bug 修复和遵循精确指令方面却无人能及。
所以,为什么不让它们合作呢?
受到 codexmcp (Python 实现) 的启发,我决定用我更喜欢的 Go 语言,边学 mcp-go-sdk 边重复造了这个轮子,主要目的是练手并打造一个更适合我自己的工具。
现在,你可以用一句话,在任何支持 MCP 的 Vibe Coding 工具中,让 Gemini 或 Claude 这样的“总指挥”去调用 Codex 这个“特种兵”来完成最棘手的编码任务。
主要特性:
SESSION_ID 维持多轮对话上下文。read-only、workspace-write 等安全策略。本工具依赖 OpenAI 的 codex CLI。请确保您已安装并配置好它。
安装 Codex CLI:
# 使用 npm 安装 (推荐)
npm i -g @openai/codex
无需安装 Go 环境,直接运行:
npx @zenfun510/codex-mcp-go
从 Releases 页面下载对应平台的二进制文件。
需要 Go 1.24+ 环境。
git clone https://github.com/w31r4/codex-mcp-go.git
cd codex-mcp-go go build -o codex-mcp-go cmd/server/main.go
根据您使用的 AI 客户端,选择对应的配置方式。
Claude Code
claude mcp add codex -s user --transport stdio -- npx -y @zenfun510/codex-mcp-go
Roo Code (VSCode / Cursor)
在 Roo Code 的 MCP 设置中添加:
{
"mcpServers": { "codex": { "command": "npx", "args": ["-y", "@zenfun510/codex-mcp-go"], "env": { "OPENAI_API_KEY": "your-api-key" } } } }
配置文件路径参考:
~/.config/Code/User/globalStorage/rooveterinaryinc.roo-cline/settings/cline_mcp_settings.json~/.config/Cursor/User/globalStorage/rooveterinaryinc.roo-cline/settings/cline_mcp_settings.jsonKiloCode
在 ~/.kilocode/mcp.json 中添加:
{
"mcpServers": { "codex": { "command": "npx", "args": ["-y", "@zenfun510/codex-mcp-go"], "env": { "OPENAI_API_KEY": "your-api-key" } } } }
</details>
Cursor (Native MCP)
codexstdionpx-y @zenfun510/codex-mcp-go如果您已通过 go build 构建了二进制文件(假设路径为 /path/to/codex-mcp-go),可使用以下配置:
Claude Code
claude mcp add codex -s user --transport stdio -- /path/to/codex-mcp-go
Roo Code / KiloCode / 通用 JSON 配置
{
"mcpServers": { "codex": { "command": "/path/to/codex-mcp-go", "args": [], "env": { "OPENAI_API_KEY": "your-api-key" } } } }
</details>
Cursor (Native MCP)
codexstdio/path/to/codex-mcp-gocat <<'EOF' | ./codex-mcp-go
{"jsonrpc":"2.0","id":1,"method":"initialize","params":{"protocolVersion":"0.1.0","capabilities":{}}} {"jsonrpc":"2.0","id":2,"method":"tools/list","params":{}} EOF
需先完成 `initialize` 握手,然后才能调用 `tools/list`。若返回包含 `codex` 工具的 JSON 数据,即表示运行正常。
支持通过 TOML 配置文件 或 环境变量 调整运行策略。优先级:默认值 < 配置文件 < 环境变量。
./codex-mcp-go --config ./codex-mcp.example.toml
CODEX_MCP_CONFIG=./codex-mcp.example.toml ./codex-mcp-go
环境变量(会覆盖配置文件):
CODEX_MCP_SERVER_NAME / CODEX_MCP_VERSIONCODEX_DEFAULT_TIMEOUT / CODEX_MAX_TIMEOUT / CODEX_NO_OUTPUT_TIMEOUT(单位:秒)CODEX_MAX_BUFFERED_LINES / CODEX_EXECUTABLE_PATHCODEX_ALLOWED_MODELS / CODEX_ALLOWED_PROFILES(逗号分隔;* 表示允许任意值;默认空=全部拒绝)CODEX_DEFAULT_SANDBOX / CODEX_ALLOWED_SANDBOX_MODES(逗号分隔)CODEX_ALLOWED_WORK_DIRS(逗号分隔的目录前缀列表;空=不限制)CODEX_DISABLE_YOLO(true/false)CODEX_LOG_LEVEL / CODEX_LOG_FORMAT / CODEX_LOG_OUTPUT / CODEX_LOG_FILE为了获得最佳体验,建议根据您使用的客户端类型配置相应的系统提示词。
适用于能够自主规划和执行多步任务的 Agent。
对于 KiloCode / Roo Code / Cline 用户: 本项目提供了针对不同客户端的预配置专家模式文件。请根据您使用的客户端选择对应的文件导入:
codex-engineer-kilocode.yamlcodex-engineer-roocode.yamlcodex-engineer-cline.yaml导入后,您将获得经过调优的 "Codex 协作专家" 模式,该模式已针对您的客户端进行了身份认同适配。
对于 Claude Code 或手动配置: 请将以下内容添加到您的 Agent 配置或作为任务的初始指令:
点击展开详细提示词
## Core Instruction for CodeX MCP
在任何时刻,你必须思考当前过程可以如何与codex进行协作,如何调用Codex 为你提供的MCP工具作为你客观全面分析的保障。 其中你务必执行以下几个步骤: 1 在你对用户需求形成初步分析后,将用户需求、初始思路告知codex,并要求其完善需求分析和实施计划。 2 在实施具体编码任务前,必须向codex索要代码实现原型(要求codex仅给出unified diff patch,严禁对代码做任何真实修改)。在获取代码原型后,你只能以此为逻辑参考,再次对代码修改进行重写,形成企业生产级别、可读性极高、可维护性极高的代码后,才能实施具体编程修改任务。 3 无论何时,只要完成切实编码行为后,必须立即使用codex review代码改动和对应需求完成程度。 4 codex只能给出参考,你必须有自己的思考,甚至需要对codex的回答提出置疑。尽信书则不如无书,你与codex的最终使命都是达成统一、全面、精准的意见,所以你们必须不断争辩已找到通向真理的唯一途径。
codex,用于执行 AI 辅助的编码任务。该工具通过 MCP 协议调用,无需使用命令行。可选参数:
~/.codex/config.toml 加载的配置文件名称,默认为 None(使用用户默认配置)推荐用法:
</details>
适用于作为 IDE 插件运行的助手。建议添加到 .clinerules (Roo Code) 或 "Rules for AI" (Cursor) 中:
点击展开规则提示词
# Codex MCP Tool Rules
You have access to the codex tool, which wraps the OpenAI Codex CLI. Use it for complex code generation, debugging, and analysis.
unified diff patch to solve the problem.sandbox="read-only" when asking for code.SESSION_ID for multi-turn tasks.cd is set to the current workspace root.sandbox="read-only". Only use workspace-write if explicitly instructed by the user and you are confident in the operation.</details>
工具名称:codex
| 参数 | 类型 | 必填 | 默认值 | 说明 |
|---|---|---|---|---|
PROMPT | string | ✅ | - | 发送给 Codex 的指令 |
cd | string | ✅ | - | 工作目录路径 |
sandbox | string | ❌ | "read-only" | 策略:read-only / workspace-write / danger-full-access |
SESSION_ID | string | ❌ | "" | 会话 ID,用于多轮对话 |
skip_git_repo_check | bool | ❌ | true | 允许在非 Git 目录运行 |
return_all_messages | bool | ❌ | false | 返回完整推理日志 |
image | []string | ❌ | [] | 附加图片路径 |
model | string | ❌ | "" | 默认禁止,除非显式允许 |
yolo | bool | ❌ | false | 跳过所有确认(非交互) |
profile | string | ❌ | "" | 默认禁止,除非显式允许 |
timeout_seconds | int | ❌ | 1800 | Codex 调用的总超时(秒,最多 1800) |
no_output_seconds | int | ❌ | 0 | 无输出达到该秒数后终止运行(0 表示关闭) |
运行时行为: 默认 30 分钟总超时(上限 30 分钟),无输出看门狗默认关闭;出现错误行、非零退出会携带最近输出返回,便于定位卡住原因。若网络慢或 MCP 客户端自身有较短的 RPC 超时,调用时保持 timeout_seconds=1800,以避免过早被取消。 默认策略: sandbox=read-only、yolo=false、skip_git_repo_check=false;model/profile 默认拒绝,需显式放行;timeout_seconds=1800(最多 1800)、no_output_seconds=0(关闭)。
| 特性 | 官方 Codex CLI | CodexMCP (本工具) |
|---|---|---|
| 基本 Codex 调用 | ✅ | ✅ |
| 多轮对话 | ❌ | ✅ (通过 Session 管理) |
| 推理详情追踪 | ❌ | ✅ (完整日志捕获) |
| 并行任务支持 | ❌ | ✅ (MCP 协议支持) |
| 错误处理 | ❌ | ✅ (结构化错误返回) |
| 特性 | Go 版本 (codex-mcp-go) | Python 版本 (codexmcp) |
|---|---|---|
| 部署 | 单二进制文件,零依赖 | 需 Python 环境及依赖 |
| 启动速度 | 🚀 极快 | 🐢 较慢 (解释器启动) |
| 资源占用 | 📉 低 | 📈 较高 |
| 并发模型 | Goroutine (高效) | Threading |
| 适用场景 | 生产环境、底层服务 | 二次开发、原型验证 |
codex CLI 是否在 PATH 中,或确认 Go 版本 >= 1.24。chmod +x)。SESSION_ID。本项目采用 MIT License 开源协议。
本项目深受 codexmcp (Python 实现) 的启发。感谢 GuDaStudio 团队在探索 Codex MCP 集成方面所做的开创性工作。
再次感谢您的关注!别忘了点个 Star 哦~ 🌟