Sequential Thinking MCP Server

What it does

The sequential thinking server provides a single structured tool for working through a problem in numbered steps. Each call carries the current thought, its number, an estimate of total thoughts, and whether another step is needed.

The tool supports revising an earlier thought, branching from a thought into an alternative path, and increasing the total thought count when the scope grows. The README positions it for planning, design, debugging and analysis where the full scope is not clear at the start.

It runs with npx or Docker. Thought logging can be turned off by setting DISABLE_THOUGHT_LOGGING to true.

Configuration

Run with npx -y @modelcontextprotocol/server-sequential-thinking, or use the mcp/sequentialthinking Docker image.

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

Source: https://github.com/modelcontextprotocol/servers/tree/main/src/sequentialthinking. Placeholders only. Replace them in your own environment, and never commit real keys to a repository.

Verified tools and features

• sequential_thinking tool Config checked

  Records one thinking step with thought, thoughtNumber, totalThoughts and nextThoughtNeeded fields.

• Revision support Config checked

  Marks a step as a revision with isRevision and revisesThought so earlier reasoning can be corrected.

• Branching support Config checked

  Branches from a prior thought using branchFromThought and branchId to explore alternative reasoning paths.

• Dynamic scope Config checked

  Adjusts totalThoughts and needsMoreThoughts as a problem grows during analysis.

Security notes

Stacks featuring this server