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)
Interactive LeetCode MCP server with authorization and solution submission capabilities
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 …”。
打开 Claude Desktop → Settings → Developer → Edit Config,把下面的内容粘贴到 claude_desktop_config.json
{
"mcpServers": {
"interactive-leetcode": {
"command": "npx",
"args": [
"-y",
"@sperekrestova/interactive-leetcode-mcp"
]
}
}
}在终端中运行这条命令
claude mcp add interactive-leetcode -- npx -y @sperekrestova/interactive-leetcode-mcp粘贴到项目中的 .cursor/mcp.json,或 ~/.cursor/mcp.json
{
"mcpServers": {
"interactive-leetcode": {
"command": "npx",
"args": [
"-y",
"@sperekrestova/interactive-leetcode-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": {
"interactive-leetcode": {
"type": "stdio",
"command": "npx",
"args": [
"-y",
"@sperekrestova/interactive-leetcode-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。
这个服务器在你的电脑上运行,而 ChatGPT 只能连接互联网上的服务器。请选择其他应用。
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。
它在你的电脑上运行;安装后,工具列表会显示在你的 AI 应用中。
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 GitHub 账户 SPerekrestova 验证,该账户拥有此仓库
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
它在你的电脑上运行,没有互联网端点
在你的电脑上运行,改用本地设置
它在你的电脑上运行;我们只能读取互联网端点的工具列表
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 1 项适用的基础检查中通过了 1 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。
它在你的电脑上运行。你的 AI 应用会在需要时启动它,数据不会经过作者的服务器(除非它自己调用在线服务)。
GitHub 仓库最后更新于–,最新版本(3.2.6)于2个月前进入注册中心。
支持 MCP 的应用,例如 Claude Desktop、Claude Code、Cursor 和 VS Code。你的电脑上还需要有 Node.js。
同一分类中有:Spec Driven Development, Remote Desktop Commander, Rystic
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.
Start, seed, clock, inspect and diff Rystic API simulators from your agent.
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。
Current project is under active development and may not work perfectly
npm install -g @sperekrestova/interactive-leetcode-mcp
interactive-leetcode-mcp --version interactive-leetcode-mcp --help
You can also run the package without installing it globally:
npx -y @sperekrestova/interactive-leetcode-mcp@latest --version
git clone https://github.com/SPerekrestova/interactive-leetcode-mcp.git
cd interactive-leetcode-mcp npm install && npm run build npm link
Use this to verify the published npm package from a clean directory without any LeetCode credentials:
mkdir leetcode-mcp-smoke
cd leetcode-mcp-smoke npm init -y npm install @sperekrestova/interactive-leetcode-mcp npx --no-install interactive-leetcode-mcp --version npx --no-install interactive-leetcode-mcp --help
Those commands confirm that Node can install the package and expose the CLI binary. For actual MCP server validation, use an integration test that connects over stdio and completes the MCP handshake.
You can execute this command in CLI
claude mcp add --transport stdio leetcode -- npx -y @sperekrestova/interactive-leetcode-mcp@latest
Or add to your MCP configuration file (~/.config/claude-code/mcp.json) or (~/Library/Application Support/Claude/claude_desktop_config.json):
{
"mcpServers": { "leetcode": { "command": "npx", "args": ["-y", "@sperekrestova/interactive-leetcode-mcp@latest"] } } }
{
"mcpServers": { "leetcode": { "command": "node", "args": ["/path/to/this/project/interactive-leetcode-mcp/build/index.js"] } } }
You: "Authorize with LeetCode"
Claude: [Opens LeetCode in your browser and guides you through the process] Claude: "Please log in to your account. Once logged in, I'll walk you through getting two cookie values we need. First, press F12 to open DevTools..." You: [Follows Claude's step-by-step guidance] You: "Here are my cookies: csrftoken is abc123... and LEETCODE_SESSION is xyz789..." Claude: "✓ Perfect! Your credentials are validated and saved. Welcome back, johndoe!"
You: "I want to practice two-sum"
Claude: [Fetches problem and creates working file]
You: "Give me a hint"
Claude: [Provides contextual guidance based on your code]
You: "Submit my solution"
Claude: "🎉 Accepted! Runtime: 2ms (beats 95.3%)"
`get_started`
`start_leetcode_auth`
`save_leetcode_credentials`
csrftoken, session (cookie values you provide)`check_auth_status`
`get_daily_challenge`
`get_problem`
titleSlug (e.g., "two-sum")`search_problems`
`start_problem`
titleSlug, optional language`request_hint`
titleSlug`get_session_state`
titleSlug`reset_session`
titleSlug`runner_doctor`
`run_local_tests`
titleSlug, language, code, optional timeoutMs`list_problem_solutions`
questionSlug, optional limit, skip, orderBy, userInput, tagSlugs`get_problem_solution`
topicId, titleSlug`submit_solution`
problemSlug, code, language`get_user_profile`
`get_recent_submissions`
`get_recent_ac_submissions`
`get_user_status`
`get_problem_submission_report`
`get_problem_progress`
`get_all_submissions`
`get_user_contest_ranking`
The Interactive LeetCode MCP includes AI agent guidance through MCP Prompts to create a better learning experience.
Workspace Setup: When learning mode is active, Claude will:
{problem-slug}.{extension}Learning-Guided Mode: When active, Claude follows these guidelines:
Problem Workflow: Guides you through the complete cycle:
To activate learning mode, tell Claude you want to practice with guidance — for example, "Let's practice in learning mode" or "I want to learn two-sum with hints." Once active:
get_daily_challenge, search_problems, or get_problemstart_problemrequest_hint rather than jumping straight to solutionsTo test the MCP server as a black-box MCP client without live LeetCode traffic:
npm run dogfood:local
This builds the server, spawns build/index.js over stdio, connects with the MCP SDK client, uses an isolated HOME, serves LeetCode responses from fixtures, and drives a user-like flow through runner_doctor, start_problem, request_hint, run_local_tests, and get_session_state.
See DOGFOOD_TESTING.md for the full workflow and a copy-paste prompt for local Claude/agent dogfood testing.
"Not authorized" or "Invalid credentials" error
"Credentials have expired"
Can't find DevTools or cookies
Copied wrong values
Browser doesn't open during authorization
"Unsupported language" error
Submission timeout
This repo also ships an agent skill that teaches Claude (and other AI agents) how to use the MCP server correctly — including session flow, prompt invocations, learning mode, and authentication.
Install the skill directly as a Claude Code plugin:
/plugin marketplace add SPerekrestova/interactive-leetcode-mcp
/plugin install interactive-leetcode-mcp@interactive-leetcode-mcp
Then start a practice session with:
/interactive-leetcode-mcp:interactive-leetcode-mcp
The skill is also published on [ClawHub](https://clawhub.ai/SPerekrestova/interact
只显示了开头部分;其余内容请在仓库中阅读