mcpField
H
huuthangntk-claude-vision-mcp-server

huuthangntk-claude-vision-mcp-server

Remote
github.com/huuthangntkAI & Agents0Updated 55w ago
Repository

About huuthangntk-claude-vision-mcp-server

Claude Deep Think MCP Server provides a `claude_think` analysis tool for developers using Anthropic Claude Sonnet 4.5. It is intended to help assess new requirements, errors, and technical decisions before code is written.

Key features

  1. 01Analyze new information before implementation
  2. 02Identify potential pitfalls and alternatives
  3. 03Offer technical and architectural guidance
  4. 04Configure the Claude model with an environment variable
  5. 05Provide Cursor rules for proactive tool use

Use cases

  1. 01Assess an error message before changing code
  2. 02Evaluate requirements before implementing a feature
  3. 03Consider architecture options for an application
  4. 04Review performance or security concerns

MCP support

Exposes the `claude_think` analysis tool to MCP-compatible clients such as Cursor IDE and Claude Desktop. The server uses an Anthropic API key and can be configured with a Claude model.

Configuration

Remote servers need no local install. Pick your client, then copy the config or install in one click.

Endpointhttps://server.smithery.ai/@huuthangntk/claude-vision-mcp-server/mcp
Copy install command
Claude CodeTerminal
Cursor.cursor/mcp.json
VS Code.vscode/mcp.json
Claude Desktopclaude_desktop_config.json
Windsurf~/.codeium/windsurf/mcp_config.json

Run commands in a terminal; paste JSON into the client’s MCP config and restart. Servers that need sign-in will prompt on first connect.

All endpoints
Streamable HTTPhttps://server.smithery.ai/@huuthangntk/claude-vision-mcp-server/mcpAuth required

Tools

Details

From the project README

Claude Deep Think MCP Server

A powerful Model Context Protocol (MCP) server that provides proactive deep analytical thinking using Anthropic's Claude Sonnet 4.5. This tool is designed to be called BEFORE writing code when new information arrives.

⚡ Core Concept: Think Before Code

Use this tool FIRST when new information arrives, BEFORE writing any code:

  • 🐛 Error messages or stack traces
  • 📝 User requirements or feature requests
  • 💬 Code review feedback
  • 🚀 Performance issues
  • 🔒 Security alerts
  • 📚 API documentation to integrate
  • 🗄️ Database problems
  • 💭 UX/UI feedback
  • 🔄 Breaking changes in dependencies
  • 🏗️ Architectural decisions

Workflow: New Info → Think Tool → Review Insights → Write Better Code

🌟 Features

Deep Think & Analysis (claude_think)

Provides intelligent insights, suggestions, and strategic guidance before code implementation. Perfect for:

  • ✅ Understanding context deeply before acting
  • ✅ Identifying potential pitfalls upfront
  • ✅ Suggesting best practices from the start
  • ✅ Offering alternative approaches
  • ✅ Extracting key information for efficient implementation
  • ✅ Strategic decision-making
  • ✅ Problem-solving and architecture planning

Result: Fewer bugs, better code quality, faster development!

📋 Prerequisites

  • Node.js 18+ or Bun
  • Anthropic Claude API key (Get one here)
  • MCP-compatible client (Cursor IDE, Claude Desktop, etc.)

🚀 Quick Start

1. Installation
cd claude-vision-mcp
bun install
# or
npm install
2. Configuration

The API key is configured when connecting to the MCP server (see Docker or Cursor setup below).

3. Build
bun run build
# or
npm run build

🐳 Docker Setup (Recommended)

Quick Start
cd claude-vision-mcp

# Create .env file with your API key
echo "ANTHROPIC_API_KEY=your-key-here" > .env
echo "CLAUDE_MODEL=claude-sonnet-4-20250514" >> .env

# Start container
docker-compose up -d

# Check status
docker ps | grep claude-vision

The container will auto-restart when Docker Desktop launches.

Docker Configuration

The server runs on http://localhost:8080/mcp with the following environment variables:

  • ANTHROPIC_API_KEY - Your Claude API key (required)
  • CLAUDE_MODEL - Model to use (default: claude-sonnet-4-20250514)

🔧 Usage in Cursor IDE

Docker Connection (Recommended)

Add to your ~/.cursor/mcp.json or .cursor/mcp.json:

{
  "mcpServers": {
    "Claude Deep Think": {
      "url": "http://localhost:8080/mcp?apiKey=YOUR_API_KEY&model=claude-sonnet-4-5-20250929"
    }
  }
}
Enable Proactive Thinking

Copy the .cursorrules file from this repo to your project root. This makes Cursor AI automatically use the think tool before writing code.

# From your project directory
cp claude-vision-mcp/.cursorrules .cursorrules
Tool Usage Pattern

Always use this pattern when new information arrives:

Use the claude_think tool to analyze: [NEW INFORMATION]

Context: [Current situation, tech stack, constraints]

Examples:

Error Message:

Use the claude_think tool to analyze:

Error: "TypeError: Cannot read property 'map' of undefined"

Context: React component rendering users from useState hook

New Feature:

Use the claude_think tool:

Requirement: Add dark mode toggle to header

Context: Next.js 14, need to check if ThemeContext exists

Performance Issue:

Use the claude_think tool:

Issue: Homepage renders 50+ times, parent causing all children to re-render

Context: useState for theme in Header, passed via props to 20+ children

📚 Examples

Example 1: Analyzing Technical Decisions
Use the claude_think tool to analyze:

"I'm building a real-time chat application. Should I use WebSockets, SSE, or HTTP polling?"

Context: Need to support 100K concurrent users, prioritize ease of implementation

Expected Output: Comprehensive comparison with pros, cons, and recommendations

Example 2: Architecture Planning
Use the claude_think tool to evaluate:

"What's the best way to structure a multi-tenant SaaS application?"

Context: PostgreSQL database, Node.js backend, 50-100 tenants expected
Example 3: Best Practices
Use the claude_think tool:

"Review this approach to handling user sessions in a Next.js app"

Context: Using JWT tokens, storing in localStorage, concerned about security

🛠️ Development

Project Structure
claude-vision-mcp/
├── src/
│   └── index.ts              # Main MCP server implementation
├── .smithery/
│   └── index.cjs            # Built server (generated)
├── package.json              # Dependencies and scripts
├── tsconfig.json            # TypeScript configuration
├── smithery.yaml            # Smithery deployment config
├── Dockerfile               # Docker container definition
├── docker-compose.yml       # Docker Compose configuration
└── README.md               # This file
Available Scripts
  • bun run build / npm run build - Compile TypeScript
  • bun run dev / npm run dev - Development server with hot reload

🔒 Security Best Practices

  1. Never commit API keys - Always use environment variables
  2. Use .gitignore - Ensure .env files are ignored
  3. Rotate keys regularly - Update API keys periodically
  4. Review tool calls - Keep manual approval enabled in Cursor
  5. Use development environments - Test with non-production data

📦 Docker Management

# Start container
docker-compose up -d

# View logs
docker logs claude-vision-mcp-server -f

# Restart container
docker-compose restart

# Stop container
docker-compose down

# Rebuild and restart
docker-compose up -d --build

🐛 Troubleshooting

Issue: Server not connecting in Cursor

Solutions:

  1. Verify Docker container is running: docker ps | grep claude-vision
  2. Check container logs: docker logs claude-vision-mcp-server
  3. Restart Cursor IDE completely
  4. Verify API key in URL is correct
Issue: API key errors

Solutions:

  1. Ensure key starts with sk-ant-
  2. Test key at: https://console.anthropic.com/
  3. Check environment variables in container
  4. Verify URL parameter format
Issue: Container won't start

Solutions:

# Check logs
docker logs claude-vision-mcp-server

# Verify .env file
cat .env

# Rebuild from scratch
docker-compose down -v
docker-compose up -d --build

💡 Performance

With Bun runtime:

  • ⚡ 4x faster package installs
  • ⚡ 3-4x faster script execution
  • 📦 Smaller Docker images
  • 🚀 Faster cold starts

📖 Comprehensive Guides

💡 Why This Workflow?

Without Think Tool:
1. User reports error
2. AI writes quick fix
3. Fix creates new bug
4. Multiple iterations needed
⏱️ Total: 30 minutes, 3 iterations
With Claude_Think Tool:
1. User reports error
2. AI analyzes with claude_think tool (20s)
3. AI writes comprehensive fix
4. Works correctly first time
⏱️ Total: 5 minutes, 1 iteration

Result: 6x faster, better quality, fewer bugs! 🎉

📄 License

MIT

🤝 Contributing

Contributions are welcome! Please feel free to submit a Pull Request.

📞 Support

For issues or questions:

  • Open an issue on GitHub
  • Check the MCP Documentation
  • Read the workflow guides in this repository

🙏 Acknowledgments

Comments

Markdown supported

New to MCP?

From what the protocol is to installing your first server, these answers cover most first questions.

An open protocol that defines how AI applications connect to external tools and data in a uniform way. Clients talk to the model; MCP servers provide tools, resources and prompt templates. They exchange standard messages, so one server can be reused by many clients.