Back to MCP directory
publicPublicdnsLocal runtime

retrieval-augmented-thinking

通过结构化的、带有检索增强的思维过程来提升人工智能模型的能力,这些过程能够实现动态思维链、并行探索路径以及递归精化周期,从而改善推理能力。

article

README

Retrieval-Augmented Thinking MCP Server

An MCP (Model Context Protocol) server implementation that enhances AI model capabilities with structured, retrieval-augmented thinking processes. This server enables dynamic thought chains, parallel exploration paths, and recursive refinement cycles for improved reasoning and problem-solving.

Features

  • Adaptive Thought Chains: Maintains coherent reasoning flows with branching and revision capabilities
  • Iterative Hypothesis Generation: Implements validation cycles for hypothesis testing
  • Context Coherence: Preserves context across non-linear reasoning paths
  • Dynamic Scope Adjustment: Supports flexible exploration and refinement
  • Quality Assessment: Real-time evaluation of thought processes
  • Branch Management: Handles parallel exploration paths
  • Revision Tracking: Manages recursive refinement cycles

Installation

npm install @modelcontextprotocol/server-retrieval-augmented-thinking

Usage

Command Line

mcp-server-retrieval-augmented-thinking

Programmatic Usage

import { Server } from '@modelcontextprotocol/sdk/server';
import { StdioServerTransport } from '@modelcontextprotocol/sdk/server/stdio';

// Initialize and run the server
const server = new Server({
  name: 'retrieval-augmented-thinking',
  version: '0.1.0'
});

// Connect transport
const transport = new StdioServerTransport();
await server.connect(transport);

Tool Configuration

The server provides a tool with the following parameters:

  • thought (string): Current reasoning step
  • thoughtNumber (number): Position in reasoning chain
  • totalThoughts (number): Estimated scope
  • nextThoughtNeeded (boolean): Chain continuation signal
  • isRevision (boolean, optional): Marks refinement steps
  • revisesThought (number, optional): References target thought
  • branchFromThought (number, optional): Branch origin point
  • branchId (string, optional): Branch identifier
  • needsMoreThoughts (boolean, optional): Scope expansion signal

Advanced Features

Thought Chain Analytics

The server tracks various metrics for thought chain quality:

  • Chain effectiveness
  • Revision impact
  • Branch success rate
  • Overall quality
  • Individual thought metrics (complexity, depth, quality, impact)

Pattern Recognition

Analyzes thought patterns for:

  • Reasoning structures
  • Context preservation
  • Hypothesis validation
  • Solution coherence

Development

# Build
npm run build

# Watch mode
npm run watch

Contributing

Contributions welcome! Please read our contributing guidelines and submit pull requests.

License

MIT

help

Runtime guide

cloud

Hosted runtime

Hosted servers run from a provider-managed environment. You usually connect the MCP client to the hosted endpoint or follow the provider's authorization flow, without keeping a local process alive

  1. Open provider connection page
  2. Authorize or copy endpoint
  3. Connect from your MCP client
terminal

Local runtime / other methods

Local servers run on your own machine or infrastructure. You normally copy the server_config into your MCP client, install the required package, and provide env variables from env_schema when needed

  1. Copy server_config
  2. Install required package
  3. Fill env variables and restart client