think-tool
Think Tool
Documentation
MCP Reflection Tool
A Model Context Protocol (MCP) server that provides a "reflect" tool for AI assistants to create cognitive checkpoints and structured reasoning. This tool helps LLMs maintain context, reflect on their work, and think through complex problems systematically.
Why Use This Tool?
Key Insight: Without explicitly outputting thought processes, no deep thinking occurs. This tool creates mandatory cognitive checkpoints that prevent shortcuts and improve accuracy.
Features
- 🧠 Structured Reasoning: Forces AI assistants to reflect step-by-step through complex problems
- ✅ Task Validation: Creates checkpoints to verify requirements are met
- 📝 Learning Documentation: Captures discoveries and insights during problem-solving
- 🔍 Debugging Aid: Helps work through issues systematically by elimination
- 🎯 Decision Audit Trail: Creates a record of reasoning for important decisions
Installation
Quick Install via NPM
# Install globally
npm install -g mcp-reflection-tool
# Or run directly with npx (no installation needed)
npx mcp-reflection-toolIntegration with AI Tools
Claude Code
Add the server using a single command:
mcp add npx mcp-reflection-toolThis will automatically configure the server in your Claude Code settings. After running the command, restart Claude Code completely.
Cursor
Add to your Cursor configuration:
Option 1: Via Settings UI
1. Open Cursor Settings (Cmd/Ctrl + ,)
2. Search for "MCP" or navigate to Features > MCP
3. Add the reflection tool configuration
Option 2: Direct Config Edit
Edit `~/.cursor/mcp_config.json`:
{
"mcpServers": {
"reflection-tool": {
"command": "npx",
"args": ["mcp-reflection-tool"]
}
}
}Restart Cursor after making changes.
Windsurf
Add to your Windsurf MCP configuration:
Location: `~/.windsurf/mcp.json` (macOS/Linux) or `%USERPROFILE%\.windsurf\mcp.json` (Windows)
{
"mcpServers": {
"reflection-tool": {
"command": "npx",
"args": ["mcp-reflection-tool"]
}
}
}Restart Windsurf to apply changes.
Cline (VS Code Extension)
Option 1: Via VS Code Settings UI
1. Open VS Code Settings (Cmd/Ctrl + ,)
2. Search for "Cline MCP"
3. Add server configuration
Option 2: Edit settings.json
Add to your VS Code `settings.json`:
{
"cline.mcpServers": {
"reflection-tool": {
"command": "npx",
"args": ["mcp-reflection-tool"]
}
}
}Reload VS Code window after configuration.
Alternative: Manual Server Mode
The server runs via stdio by default. If you need HTTP mode, use environment variables:
# Start server in stdio mode (default)
npx mcp-reflection-tool
# Start in HTTP mode on port 8080
HTTP=true npx mcp-reflection-tool
# Start in HTTP mode on custom port
HTTP=true PORT=3000 npx mcp-reflection-toolMost modern AI tools support stdio mode automatically.
Usage Examples
Once installed, the AI assistant will have access to the `reflect` tool. Here are examples of how it gets used:
Before Complex Actions
Using reflection tool: "Breaking down authentication implementation:
1. Check existing auth patterns in codebase
2. Set up JWT token generation
3. Add middleware for route protection
4. Test with valid and expired tokens"After Completing Tasks
Using reflection tool: "Task completion check:
- Completed: Implemented user authentication with JWT
- Learned: Existing middleware made integration smooth
- Technical debt: Need to add rate limiting
- Next: Update API documentation"During Problem Solving
Using reflection tool: "Debugging slow API responses:
- Symptom: 5+ second response times
- Hypothesis 1: Missing database indexes - CONFIRMED
- Hypothesis 2: N+1 query problem - Also found
- Solution: Added compound indexes and query batching
- Result: Response time now Output > Select "Cline" from dropdown
- **Cursor logs**: Help > Toggle Developer Tools > Console
### Verifying InstallationCheck if package is installed globally
npm list -g mcp-reflection-tool
Test the server directly
npx mcp-reflection-tool
Test stdio mode
echo '{"jsonrpc":"2.0","method":"initialize","id":1}' | npx mcp-reflection-tool
## Contributing
Contributions are welcome! Please feel free to submit a Pull Request.
1. Fork the repository
2. Create your feature branch (`git checkout -b feature/amazing-feature`)
3. Commit your changes (`git commit -m 'Add amazing feature'`)
4. Push to the branch (`git push origin feature/amazing-feature`)
5. Open a Pull Request
## License
MIT License - see the LICENSE file for details.
## Author
Created as an MCP implementation for enhancing AI reasoning capabilities.
## Links
- [NPM Package](https://www.npmjs.com/package/mcp-reflection-tool)
- [GitHub Repository](https://github.com/sterling/think-tool)
- [Model Context Protocol Docs](https://modelcontextprotocol.io/)
- [Report Issues](https://github.com/sterling/think-tool/issues)
## Acknowledgments
- Built with [FastMCP](https://github.com/spencerc99/fastmcp) framework
- Implements the [Model Context Protocol](https://modelcontextprotocol.io/) standard
- Inspired by research on structured reasoning for AI systemsFrequently asked questions
What is think-tool?
think-tool is Think Tool
How do I install think-tool?
Open the GitHub repository and follow its README. Most MCP servers are added to your client's MCP config, then called by your agent.
Is think-tool open source?
Yes — it is hosted on GitHub at https://github.com/fourcolors/think-tool and has 6 stars.
Related MCP tools
serversModel Context Protocol Servers
UI-TARS-desktopThe Open-Source Multimodal AI Agent Stack: Connecting Cutting-Edge AI Models and Agent Infra
n8n-mcpA MCP for Claude Desktop / Claude Code / Windsurf / Cursor to build n8n workflows for you
Figma-Context-MCPMCP server to provide Figma layout information to AI coding agents like Cursor
inbox-zeroThe world's best AI personal assistant for email. Open source app to help you reach inbox zero fast.
instantInstant is the best backend for AI-coded apps. You get auth, permissions, storage, presence, and streams — everything you need to ship apps your users will love.
Run your own MCP server? See who uses it and what to fix.
Measure it with TrackMCP