Claude MCP Server
Exposes Anthropic's Claude as 3 MCP tools (ask_claude, claude_analyze, claude_write) that Manus can call for general prompting, structured analysis, and content generation.
README
Claude MCP Server
Connect Manus → Claude (Anthropic API) via MCP
This server exposes Claude as 3 MCP tools that Manus can call:
- ask_claude — General prompt → response
- claude_analyze — Structured analysis (summarize, critique, extract, etc.)
- claude_write — Content generation (email, report, social post, etc.)
Quick Start (Local)
# 1. Install dependencies
npm install
# 2. Set your API key
cp .env.example .env
# Edit .env and add your ANTHROPIC_API_KEY
# 3. Run in dev mode
npm run dev
# Server runs at http://localhost:3000
# MCP endpoint: http://localhost:3000/mcp
Deploy to Railway (FREE — Recommended)
Railway gives you a free hosted URL — perfect for Manus to connect to.
Steps:
- Push to GitHub
git init
git add .
git commit -m "initial commit"
gh repo create claude-mcp-server --public --push
# or use github.com to create repo manually
- Deploy on Railway
- Go to railway.app
- Click New Project → Deploy from GitHub Repo
- Select your repo
- Click Variables → Add:
ANTHROPIC_API_KEY= your key from console.anthropic.comPORT= 3000
- Railway auto-builds and gives you a URL like:
https://claude-mcp-server-production.up.railway.app
- Your MCP endpoint will be:
https://claude-mcp-server-production.up.railway.app/mcp
Deploy to Render (FREE tier available)
- Go to render.com
- Click New → Web Service
- Connect your GitHub repo
- Set:
- Build Command:
npm install && npm run build - Start Command:
npm start - Environment Variables:
ANTHROPIC_API_KEY=your-key
- Build Command:
- Deploy — you get a URL like
https://claude-mcp-server.onrender.com
⚠️ Render free tier spins down after 15 mins of inactivity (cold start ~30s). Railway free tier stays live — better for Manus.
Deploy to Fly.io (FREE hobby tier)
# Install flyctl
curl -L https://fly.io/install.sh | sh
# Login
fly auth login
# Launch
fly launch
# Set secret
fly secrets set ANTHROPIC_API_KEY=sk-ant-your-key-here
# Deploy
fly deploy
Connect to Manus
- In Manus go to Settings → Integrations → Custom MCP Servers
- Click Add Server
- Paste your URL:
https://your-app.up.railway.app/mcp - Save — Manus will now show 3 Claude tools available
Example Manus prompts once connected:
- "Use ask_claude to write a professional bio for me"
- "Use claude_analyze to summarize this document [paste text]"
- "Use claude_write to draft a LinkedIn post about X in a casual tone"
Tools Reference
ask_claude
| Parameter | Type | Required | Description |
|---|---|---|---|
| prompt | string | ✅ | The question or task for Claude |
| system | string | ❌ | Optional system prompt / persona |
| max_tokens | number | ❌ | Max response length (default: 1024) |
claude_analyze
| Parameter | Type | Required | Description |
|---|---|---|---|
| content | string | ✅ | Text to analyze |
| focus | enum | ✅ | summarize / critique / compare / extract_key_points / classify |
| instructions | string | ❌ | Extra instructions |
claude_write
| Parameter | Type | Required | Description |
|---|---|---|---|
| task | string | ✅ | What to write |
| format | enum | ✅ | email / report / social_post / bullet_list / paragraph / code |
| tone | enum | ❌ | professional / casual / persuasive / empathetic / technical |
| context | string | ❌ | Background context |
Health Check
curl https://your-app.up.railway.app/health
# → {"status":"ok","server":"claude-mcp-server"}
Cost
- Hosting: Free on Railway (500 hrs/month free) or Render
- API: Anthropic charges per token used — Claude Sonnet is ~$3/million input tokens
- A typical ask_claude call costs < $0.01
推荐服务器
Baidu Map
百度地图核心API现已全面兼容MCP协议,是国内首家兼容MCP协议的地图服务商。
Playwright MCP Server
一个模型上下文协议服务器,它使大型语言模型能够通过结构化的可访问性快照与网页进行交互,而无需视觉模型或屏幕截图。
Magic Component Platform (MCP)
一个由人工智能驱动的工具,可以从自然语言描述生成现代化的用户界面组件,并与流行的集成开发环境(IDE)集成,从而简化用户界面开发流程。
Audiense Insights MCP Server
通过模型上下文协议启用与 Audiense Insights 账户的交互,从而促进营销洞察和受众数据的提取和分析,包括人口统计信息、行为和影响者互动。
VeyraX
一个单一的 MCP 工具,连接你所有喜爱的工具:Gmail、日历以及其他 40 多个工具。
graphlit-mcp-server
模型上下文协议 (MCP) 服务器实现了 MCP 客户端与 Graphlit 服务之间的集成。 除了网络爬取之外,还可以将任何内容(从 Slack 到 Gmail 再到播客订阅源)导入到 Graphlit 项目中,然后从 MCP 客户端检索相关内容。
Kagi MCP Server
一个 MCP 服务器,集成了 Kagi 搜索功能和 Claude AI,使 Claude 能够在回答需要最新信息的问题时执行实时网络搜索。
e2b-mcp-server
使用 MCP 通过 e2b 运行代码。
Neon MCP Server
用于与 Neon 管理 API 和数据库交互的 MCP 服务器
Exa MCP Server
模型上下文协议(MCP)服务器允许像 Claude 这样的 AI 助手使用 Exa AI 搜索 API 进行网络搜索。这种设置允许 AI 模型以安全和受控的方式获取实时的网络信息。