Official implementation of Anthropic's "reasoning" tool as an MCP server - Dramatically improve Claude's reasoning capabilities with structured thinking.
This MCP server implements the exact "reasoning" tool that Anthropic introduced in their engineering blog post. The Reasoning Tool provides Claude with a dedicated space for structured reasoning during complex problem-solving tasks, enabling more thoughtful, accurate, and reliable responses.
Anthropic's research demonstrates remarkable improvements when using the "reasoning" tool:
- 54% improvement in complex customer service tasks
- Significantly better adherence to detailed policies and guidelines
- Enhanced consistency across multiple trials of the same task
- Improved performance on software engineering benchmarks
- Minimal implementation overhead compared to other enhancement techniques
The "reasoning" tool excels where other approaches fall short:
- Better than extended thinking for cases requiring complex tool chains
- More effective than baseline prompting for policy-heavy scenarios
- Especially powerful when paired with optimized prompting
Quick Install
For Claude Desktop
npx -y @smithery/cli@latest install @NexusX-MCP/reasoning-mcp-server --client claude --config "{}"
For Cursor
npx -y @smithery/cli@latest install @NexusX-MCP/reasoning-mcp-server --client cursor --config "{}"
How It Works
The "reasoning" tool implements the exact mechanism described in Anthropic's engineering blog. Unlike extended thinking (which happens before Claude starts responding), the "reasoning" tool allows Claude to pause and reflect during its response generation.
Key mechanism: The tool doesn't perform any external actions or retrieve new information - it simply provides Claude with a dedicated scratchpad to work through reasoning step-by-step, which dramatically improves performance on complex tasks.
When Claude uses the "reasoning" tool:
- It pauses to organize thoughts before continuing a complex reasoning chain
- It creates a structured approach to multi-step problems
- It verifies policy compliance more thoroughly and consistently
- It carefully analyzes tool outputs before deciding next steps
- It maintains better context awareness across long interactions
The "reasoning" tool is especially valuable when:
- Working with other MCP tools - Great for analyzing outputs from databases, filesystems, or APIs
- Following complex policies - Perfect for customer service, legal, or compliance scenarios
- Making sequential decisions - Ideal for workflows where later steps depend on earlier ones
- Processing web search results - Helps Claude synthesize information from multiple sources
- Solving coding challenges - Improves success rates on software engineering tasks
System Prompt for Optimal Results
Anthropic's research shows that combining the "reasoning" tool with optimized prompting delivers the strongest performance improvements. For best results, add the following optimized system prompt to your Claude interaction:
For Claude Desktop (Custom Instructions)
- Go to Settings > Custom Instructions
- Add the following system prompt:
You have access to a "reasoning" tool that provides a dedicated space for structured reasoning. Using this tool significantly improves your performance on complex tasks.
## When to use the reasoning tool
Before taking any action or responding to the user after receiving tool results, use the reasoning tool as a scratchpad to:
- List the specific rules that apply to the current request
- Check if all required information is collected
- Verify that the planned action complies with all policies
- Iterate over tool results for correctness
- Analyze complex information from web searches or other tools
- Plan multi-step approaches before executing them
## How to use the reasoning tool effectively
When using the reasoning tool:
1. Break down complex problems into clearly defined steps
2. Identify key facts, constraints, and requirements
3. Check for gaps in information and plan how to fill them
4. Evaluate multiple approaches before choosing one
5. Verify your reasoning for logical errors or biases
Remember that using the reasoning tool has been shown to improve your performance by up to 54% on complex tasks, especially when working with multiple tools or following detailed policies.
For Cursor (Global Rules)
To add the Reasoning Tool as a Cursor Rule:
- Open Cursor Settings
- Navigate to General > Rules for AI
- Add a new rule with the following content:
After any context change (viewing new files, running commands, or receiving tool outputs), use the "mcp_reasoning" tool to organize your reasoning before responding.
Specifically, always use the reasoning tool when:
- After examining file contents or project structure
- After running terminal commands or analyzing their outputs
- After receiving search results or API responses
- Before making code suggestions or explaining complex concepts
- When transitioning between different parts of a task
When using the reasoning tool:
- List the specific rules or constraints that apply to the current task
- Check if all required information is collected
- Verify that your planned approach is correct
- Break down complex problems into clearly defined steps
- Analyze outputs from other tools thoroughly
- Plan multi-step approaches before executing them
The reasoning tool has been proven to improve performance by up to 54% on complex tasks, especially when working with multiple tools or following detailed policies.
Manual Installation
If you prefer to run the server locally:
- Clone the repository:
git clone https://github.com/NexusX-MCP/reasoning-mcp-server.git
cd reasoning-mcp-server
- Install dependencies:
- Build and run:
- Configure Claude Desktop manually:
- Find or create the configuration file:
- macOS:
~/Library/Application Support/Claude/claude_desktop_config.json - Windows:
%APPDATA%\Claude\claude_desktop_config.json
- Add your server configuration:
{
"mcpServers": {
"reasoning-tool": {
"command": "node",
"args": ["path/to/reasoning-mcp-server/dist/server.js"]
}
}
}
License
MIT License