{
  "markdown": "# Claude Deep Think MCP Server\n\nA 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.\n\n## ⚡ Core Concept: Think Before Code\n\n**Use this tool FIRST when new information arrives, BEFORE writing any code:**\n\n- 🐛 Error messages or stack traces\n- 📝 User requirements or feature requests\n- 💬 Code review feedback\n- 🚀 Performance issues\n- 🔒 Security alerts\n- 📚 API documentation to integrate\n- 🗄️ Database problems\n- 💭 UX/UI feedback\n- 🔄 Breaking changes in dependencies\n- 🏗️ Architectural decisions\n\n**Workflow**: New Info → Think Tool → Review Insights → Write Better Code\n\n## 🌟 Features\n\n### Deep Think & Analysis (`claude_think`)\nProvides intelligent insights, suggestions, and strategic guidance **before code implementation**. Perfect for:\n- ✅ Understanding context deeply before acting\n- ✅ Identifying potential pitfalls upfront\n- ✅ Suggesting best practices from the start\n- ✅ Offering alternative approaches\n- ✅ Extracting key information for efficient implementation\n- ✅ Strategic decision-making\n- ✅ Problem-solving and architecture planning\n\n**Result**: Fewer bugs, better code quality, faster development!\n\n## 📋 Prerequisites\n\n- Node.js 18+ or Bun\n- Anthropic Claude API key ([Get one here](https://console.anthropic.com/))\n- MCP-compatible client (Cursor IDE, Claude Desktop, etc.)\n\n## 🚀 Quick Start\n\n### 1. Installation\n\n```bash\ncd claude-vision-mcp\nbun install\n# or\nnpm install\n```\n\n### 2. Configuration\n\nThe API key is configured when connecting to the MCP server (see Docker or Cursor setup below).\n\n### 3. Build\n\n```bash\nbun run build\n# or\nnpm run build\n```\n\n## 🐳 Docker Setup (Recommended)\n\n### Quick Start\n\n```bash\ncd claude-vision-mcp\n\n# Create .env file with your API key\necho \"ANTHROPIC_API_KEY=your-key-here\" > .env\necho \"CLAUDE_MODEL=claude-sonnet-4-20250514\" >> .env\n\n# Start container\ndocker-compose up -d\n\n# Check status\ndocker ps | grep claude-vision\n```\n\nThe container will auto-restart when Docker Desktop launches.\n\n### Docker Configuration\n\nThe server runs on `http://localhost:8080/mcp` with the following environment variables:\n- `ANTHROPIC_API_KEY` - Your Claude API key (required)\n- `CLAUDE_MODEL` - Model to use (default: claude-sonnet-4-20250514)\n\n## 🔧 Usage in Cursor IDE\n\n### Docker Connection (Recommended)\n\nAdd to your `~/.cursor/mcp.json` or `.cursor/mcp.json`:\n\n```json\n{\n  \"mcpServers\": {\n    \"Claude Deep Think\": {\n      \"url\": \"http://localhost:8080/mcp?apiKey=YOUR_API_KEY&model=claude-sonnet-4-5-20250929\"\n    }\n  }\n}\n```\n\n### Enable Proactive Thinking\n\nCopy the `.cursorrules` file from this repo to your project root. This makes Cursor AI automatically use the think tool before writing code.\n\n```bash\n# From your project directory\ncp claude-vision-mcp/.cursorrules .cursorrules\n```\n\n### Tool Usage Pattern\n\n**Always use this pattern when new information arrives:**\n\n```\nUse the claude_think tool to analyze: [NEW INFORMATION]\n\nContext: [Current situation, tech stack, constraints]\n```\n\n**Examples:**\n\n**Error Message:**\n```\nUse the claude_think tool to analyze:\n\nError: \"TypeError: Cannot read property 'map' of undefined\"\n\nContext: React component rendering users from useState hook\n```\n\n**New Feature:**\n```\nUse the claude_think tool:\n\nRequirement: Add dark mode toggle to header\n\nContext: Next.js 14, need to check if ThemeContext exists\n```\n\n**Performance Issue:**\n```\nUse the claude_think tool:\n\nIssue: Homepage renders 50+ times, parent causing all children to re-render\n\nContext: useState for theme in Header, passed via props to 20+ children\n```\n\n## 📚 Examples\n\n### Example 1: Analyzing Technical Decisions\n\n```\nUse the claude_think tool to analyze:\n\n\"I'm building a real-time chat application. Should I use WebSockets, SSE, or HTTP polling?\"\n\nContext: Need to support 100K concurrent users, prioritize ease of implementation\n```\n\n**Expected Output**: Comprehensive comparison with pros, cons, and recommendations\n\n### Example 2: Architecture Planning\n\n```\nUse the claude_think tool to evaluate:\n\n\"What's the best way to structure a multi-tenant SaaS application?\"\n\nContext: PostgreSQL database, Node.js backend, 50-100 tenants expected\n```\n\n### Example 3: Best Practices\n\n```\nUse the claude_think tool:\n\n\"Review this approach to handling user sessions in a Next.js app\"\n\nContext: Using JWT tokens, storing in localStorage, concerned about security\n```\n\n## 🛠️ Development\n\n### Project Structure\n\n```\nclaude-vision-mcp/\n├── src/\n│   └── index.ts              # Main MCP server implementation\n├── .smithery/\n│   └── index.cjs            # Built server (generated)\n├── package.json              # Dependencies and scripts\n├── tsconfig.json            # TypeScript configuration\n├── smithery.yaml            # Smithery deployment config\n├── Dockerfile               # Docker container definition\n├── docker-compose.yml       # Docker Compose configuration\n└── README.md               # This file\n```\n\n### Available Scripts\n\n- `bun run build` / `npm run build` - Compile TypeScript\n- `bun run dev` / `npm run dev` - Development server with hot reload\n\n## 🔒 Security Best Practices\n\n1. **Never commit API keys** - Always use environment variables\n2. **Use .gitignore** - Ensure `.env` files are ignored\n3. **Rotate keys regularly** - Update API keys periodically\n4. **Review tool calls** - Keep manual approval enabled in Cursor\n5. **Use development environments** - Test with non-production data\n\n## 📦 Docker Management\n\n```bash\n# Start container\ndocker-compose up -d\n\n# View logs\ndocker logs claude-vision-mcp-server -f\n\n# Restart container\ndocker-compose restart\n\n# Stop container\ndocker-compose down\n\n# Rebuild and restart\ndocker-compose up -d --build\n```\n\n## 🐛 Troubleshooting\n\n### Issue: Server not connecting in Cursor\n\n**Solutions:**\n1. Verify Docker container is running: `docker ps | grep claude-vision`\n2. Check container logs: `docker logs claude-vision-mcp-server`\n3. Restart Cursor IDE completely\n4. Verify API key in URL is correct\n\n### Issue: API key errors\n\n**Solutions:**\n1. Ensure key starts with `sk-ant-`\n2. Test key at: https://console.anthropic.com/\n3. Check environment variables in container\n4. Verify URL parameter format\n\n### Issue: Container won't start\n\n**Solutions:**\n```bash\n# Check logs\ndocker logs claude-vision-mcp-server\n\n# Verify .env file\ncat .env\n\n# Rebuild from scratch\ndocker-compose down -v\ndocker-compose up -d --build\n```\n\n## 💡 Performance\n\nWith Bun runtime:\n- ⚡ 4x faster package installs\n- ⚡ 3-4x faster script execution  \n- 📦 Smaller Docker images\n- 🚀 Faster cold starts\n\n## 📖 Comprehensive Guides\n\n- **[PROACTIVE_THINKING_WORKFLOW.md](./PROACTIVE_THINKING_WORKFLOW.md)** - Complete workflow guide with before/after examples\n- **[THINK_TOOL_EXAMPLES.md](./THINK_TOOL_EXAMPLES.md)** - 10 real-world usage examples\n- **[.cursorrules](./.cursorrules)** - Cursor IDE rules for automatic think-before-code pattern\n\n## 💡 Why This Workflow?\n\n### Without Think Tool:\n```\n1. User reports error\n2. AI writes quick fix\n3. Fix creates new bug\n4. Multiple iterations needed\n⏱️ Total: 30 minutes, 3 iterations\n```\n\n### With Claude_Think Tool:\n```\n1. User reports error\n2. AI analyzes with claude_think tool (20s)\n3. AI writes comprehensive fix\n4. Works correctly first time\n⏱️ Total: 5 minutes, 1 iteration\n```\n\n**Result**: 6x faster, better quality, fewer bugs! 🎉\n\n## 📄 License\n\nMIT\n\n## 🤝 Contributing\n\nContributions are welcome! Please feel free to submit a Pull Request.\n\n## 📞 Support\n\nFor issues or questions:\n- Open an issue on GitHub\n- Check the [MCP Documentation](https://modelcontextprotocol.io/)\n- Read the workflow guides in this repository\n\n## 🙏 Acknowledgments\n\n- Built with [Anthropic Claude API](https://www.anthropic.com/)\n- Powered by [Model Context Protocol](https://modelcontextprotocol.io/)\n- Containerized with [Bun](https://bun.sh/)",
  "bytes": 7937,
  "sha": "daaf309eafa7de6b16e6ab1ea71494638c8271e192cb0529e96d4bbb4240f271",
  "repo_slug": "huuthangntk/claude-vision-mcp-server",
  "fonte": "repo",
  "truncated": false,
  "api": "https://agentalog.com/api/listings/mcp_ai_smithery_huuthangntk_claude_vision_mc_13ff51f3/readme"
}