Spec Driven Development
MCP Server that facilitates spec-driven development workflows, not just Vibe Coding.
作者提供的说明(英文原文)
Connect accounts once. Set what agents may do per account. Enforced when the action runs.
这三项永远不会合并成一个分数:意见、测量和人气是不同的东西。
我们还没有为它写通俗的介绍。下面是我们从注册中心读取到的事实。
选择你使用的 AI 应用,然后复制设置。
你需要
你用的是哪个应用?
在上方选择你的应用。我们会记住它,下次按钮会显示“安装到 …”。
适用于网页版、桌面版和手机版 Claude:
https://app.kinlab.ai/mcp在终端中运行这条命令
claude mcp add --transport http kin https://app.kinlab.ai/mcp粘贴到项目中的 .cursor/mcp.json,或 ~/.cursor/mcp.json
{
"mcpServers": {
"kin": {
"url": "https://app.kinlab.ai/mcp"
}
}
}浏览器会先询问是否打开应用,然后应用会显示设置并请你确认。密钥的值会留空,由你自己填写。
粘贴到项目中的 .vscode/mcp.json
{
"servers": {
"kin": {
"type": "http",
"url": "https://app.kinlab.ai/mcp"
}
}
}浏览器会先询问是否打开应用,然后应用会显示设置并请你确认。密钥的值会留空,由你自己填写。
ChatGPT 只能连接互联网上的服务器:
https://app.kinlab.ai/mcp菜单名称可能因套餐和版本不同而略有差异。
你的选择通过 Cookie 保存在这台设备上,无需账户。
这些设置是根据作者在注册中心的条目生成的。如果 README 的写法不同,请以 README 为准。
服务器要求先登录,才会列出它的工具。
我们检查的是任何人都能看到的信息;我们没有阅读代码。不适用的检查项不计入总数。 · 检查于 2026年11月7日
注册名称已通过域名 kinlab.ai 验证(注册中心会核实域名所有权)
我们还没有从 GitHub 读取这个仓库
我们还没有从 GitHub 读取这个仓库
在 108 毫秒内响应(HTTP 401)
声明支持 OAuth 登录,你的 AI 应用可以带你去登录
尚未检查;我们会在检查端点时读取工具列表
与注册中心里这个分类中所有活跃的服务器比较。只有当发布者拥有该仓库时,星标数才计入。
它在 3 项适用的基础检查中通过了 3 项。我们没有阅读代码。在添加密钥或令牌之前,请先看看这个服务器要求哪些权限,并使用只具备所需权限的密钥。
注册中心没有列出必需的密钥,不过有些服务器会在首次使用时要求你登录。
它在作者的云端运行。通过互联网连接,无需安装,但你发送的内容会经过作者的服务器。
GitHub 仓库最后更新于–,最新版本(0.9.1)于6天前进入注册中心。
支持 MCP 的应用,例如 Claude Desktop、Claude Code、Cursor 和 VS Code。它通过 URL 连接,所以接受远程 MCP 服务器的应用也可以使用,例如开启开发者模式的 ChatGPT。
同一分类中有:Spec Driven Development, Remote Desktop Commander, FlyDocs
同一分类中其他作者的服务器,按星标数排序。
如何看卡片:★ GitHub 星标i三大支柱:评价 · 检查 · 使用情况i更新 = 最新版本发布到注册中心的时间i
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.
Spec-driven development workflow: your agent works issues, sessions and the status lifecycle.
按原文从作者的仓库复制。PopNx MCP 没有审查或背书其内容。HTML、图片和脚本已被移除。链接会离开本站。
Kin lets your AI apps use your accounts, even two of one service, each with its own rules. Connect a work GitHub and a personal GitHub, a work Notion and a personal one. For each account and action, choose Allow, Ask me or Block. Kin checks the rule on its server when the action runs, in every AI app you connect.
This repository holds no server code. Kin runs as a hosted MCP server at https://app.kinlab.ai/mcp. What's here is the setup that points each AI app at that address, two skills that teach the app how to use Kin's tools, and an optional Claude Code hook. You need a Kin account, free at kinlab.ai, and an account at each service you connect.
Claude Code:
claude plugin marketplace add kinlab-ai/kin-plugin
claude plugin install kin@kin
Then run /mcp, choose kin and sign in. To send Claude Code's permission prompts to Kin as well, add claude plugin install kin-approvals@kin (see below).
Codex:
codex plugin marketplace add kinlab-ai/kin-plugin
codex plugin add kin@kin
Gemini CLI:
gemini extensions install https://github.com/kinlab-ai/kin-plugin
Antigravity:
git clone https://github.com/kinlab-ai/kin-plugin
agy plugin install kin-plugin/antigravity
Then select Authenticate next to kin_kin in Agent settings, Customizations.
Cursor, VS Code, LM Studio and Kiro: open the link for your app, then sign in.
Add to Cursor · Add to VS Code · Add to LM Studio · Add to Kiro
Claude, ChatGPT, Zed, Raycast, Warp and other AI apps: add https://app.kinlab.ai/mcp as a remote MCP server and sign in. The setup guide for each app has the steps.
| File | What it does |
|---|---|
.claude-plugin/plugin.json, .claude-plugin/marketplace.json | The Claude Code plugin kin and this repository's plugin list |
.mcp.json | Kin's address for Claude Code: https://app.kinlab.ai/mcp, signed in with OAuth |
plugin.json, mcp.json, .agents/plugins/marketplace.json | The same plugin and address for Codex |
gemini-extension.json, GEMINI.md | The same address and a short note for Gemini CLI |
antigravity/ | The same plugin for Antigravity, with its own manifest and copies of the skills |
server.json | Kin's entry for the MCP Registry |
skills/kin | How to call Kin's tools (search, describe, read, execute, feedback) and read their answers |
skills/kin-accounts | How to work with several accounts of one service |
approvals/ | The optional kin-approvals plugin for Claude Code, below |
assets/ | Kin's icon |
None of these files hold a key or a password. Each app signs you in to Kin through Kin's own sign-in page.
kin-approvals adds one Claude Code hook. When Claude Code is about to ask your permission for a tool, the hook runs approvals/scripts/kin-permission.sh, which hands the request to Kin's own permission hook on this computer. That hook comes with the Kin app on a Mac or with kin-bridge on Windows and Linux. It sends the request to your Kin account, so you can answer it on your phone or in Kin, and passes your answer back to Claude Code.
The script does nothing, and Claude Code asks you as usual, when the Kin app and kin-bridge aren't installed, or when Kin's hook is already in your Claude Code settings. It downloads and installs nothing, reads no credentials, and sends nothing on its own. Claude Code's deny and ask rules still apply.
When an AI app uses Kin, Kin receives the calls it makes: the action, its arguments and the reason the app gives. With the approvals hook, Kin also receives the permission request Claude Code would have shown you. Kin keeps these in your activity log. See Kin's privacy policy.
Docs: kinlab.ai/docs. Support: support@kinlab.ai. Security reports: security@kinlab.ai.