Sequential Thinking

​

Overview

​

Features

• Break complex problems into sequential steps
• Revise and refine thoughts as understanding evolves
• Branch into alternative reasoning paths
• Dynamically adjust total thought count
• Generate and verify solution hypotheses
• Maintain context across multiple thinking steps

​

Installation

​

NPX

npx -y @modelcontextprotocol/server-sequential-thinking

​

Docker

docker run --rm -i mcp/sequentialthinking

​

Configuration

​

Claude Desktop

{
  "mcpServers": {
    "sequential-thinking": {
      "command": "npx",
      "args": [
        "-y",
        "@modelcontextprotocol/server-sequential-thinking"
      ]
    }
  }
}

{
  "mcpServers": {
    "sequential-thinking": {
      "command": "npx",
      "args": [
        "-y",
        "@modelcontextprotocol/server-sequential-thinking"
      ],
      "env": {
        "DISABLE_THOUGHT_LOGGING": "true"
      }
    }
  }
}

​

VS Code

{
  "servers": {
    "sequential-thinking": {
      "command": "npx",
      "args": [
        "-y",
        "@modelcontextprotocol/server-sequential-thinking"
      ]
    }
  }
}

​

Tool

​

sequential_thinking

• thought (string, required): The current thinking step content
• nextThoughtNeeded (boolean, required): Whether another thought step is needed
• thoughtNumber (integer, required): Current thought number in sequence
• totalThoughts (integer, required): Estimated total thoughts needed
• isRevision (boolean, optional): Whether this revises previous thinking
• revisesThought (integer, optional): Which thought number is being reconsidered
• branchFromThought (integer, optional): Thought number where branching occurs
• branchId (string, optional): Identifier for the alternative reasoning branch
• needsMoreThoughts (boolean, optional): Request to extend the thought sequence

​

Use Cases

• Complex Problem Decomposition: Break multi-faceted problems into logical steps
• Iterative Planning: Design solutions with room for revision as constraints emerge
• Course-Correcting Analysis: Adjust reasoning when new information changes direction
• Uncertain Scope: Problems where full complexity isn’t clear initially
• Context Preservation: Maintain reasoning chain across extended analysis
• Information Filtering: Separate relevant from irrelevant details systematically

​

Example Usage

1. Initial Thought (thoughtNumber: 1, totalThoughts: 5)
   • State the problem and initial approach
2. Exploration (thoughtNumber: 2-4)
   • Work through the solution step by step
   • Set nextThoughtNeeded: true to continue
3. Revision (if needed)
   • Set isRevision: true and revisesThought: 2
   • Correct earlier reasoning based on new insights
4. Branching (for alternatives)
   • Set branchFromThought: 3 and branchId: "alternative-a"
   • Explore different solution path
5. Dynamic Extension
   • Set needsMoreThoughts: true when complexity exceeds estimate
   • Adjust totalThoughts upward as needed
6. Conclusion (thoughtNumber: 5, nextThoughtNeeded: false)
   • Synthesize findings and present solution

​

Environment Variables

• DISABLE_THOUGHT_LOGGING: Set to true to disable logging of thought information

​

Building

docker build -t mcp/sequentialthinking -f src/sequentialthinking/Dockerfile .