Anthropic Official AI

Sequential Thinking

Structured chain-of-thought scaffolding — the model writes its own multi-step plan as MCP calls.

85,593 stars 10,674 forks Last commit 1d ago Language TypeScript License MIT
Sign in to like + track View on GitHub →

Install

npm install -g @modelcontextprotocol/server-sequential-thinking

Claude Desktop / Cursor config

Drop into your claude_desktop_config.json or Cursor MCP settings. Replace the ${…} placeholders with your own values.

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

About

Forces the model to externalise its reasoning as a sequence of MCP tool calls. Each step has an explicit goal + result; downstream steps reference upstream ones. Used to make CoT reliable on hard problems where ad-hoc reasoning would lose context.

Discussion (0)

Sign in to comment →

No comments yet. Sign in to start the discussion.

More AI servers