Systematic Reasoning AI MCP Server

Systematic Reasoning AI MCP Server

Enforces a structured reasoning cycle with mandatory reflection and searchable memory, transforming AI agents into deliberate, learning-based thinkers.

Category
访问服务器

README

🚀 Systematic Reasoning AI: The Cognitive Engine for Intelligent Agents

<div align="center"> <img src="image/logo.png" alt="Systematic Reasoning AI Logo" width="500" />

<h1>The Cognitive Engine for Intelligent Agents</h1>

<h3>An MCP Server that transforms any AI into a deliberate, learning-based super-thinker.</h3>

<p> <a href="https://github.com/rayss868/systematic-reasoning-ai-mcp/blob/main/LICENSE"><img src="https://img.shields.io/badge/License-MIT-blue.svg" alt="License"></a> <img src="https://img.shields.io/badge/Version-1.0.0-purple.svg" alt="Version"> <img src="https://img.shields.io/badge/Status-Production%20Ready-brightgreen.svg" alt="Status"> </p> </div>


🌟 What if your AI could Evolve?

What separates a simple script from true intelligence? The ability to learn.

This project provides the missing piece. It's a plug-and-play cognitive engine that forces any AI agent to adopt the habits of a genius:

  • 🤔 Ponder every move: It mandates a "think before you act" philosophy.
  • 🧠 Never forget a lesson: It records every success and failure into a permanent, searchable memory.
  • 📖 Continuously improve: It uses past experiences to make smarter decisions in the future.

This isn't just a tool. It's an upgrade to your AI's core operating system.


🏛️ The Three Pillars of an Evolved AI

<div align="center"> <img src="https://media.giphy.com/media/v1.Y2lkPTc5MGI3NjExbDB6d2Q0d2ZidmN1N2N6c2JjZ3g0dGxlNXQyN3A2Z3kzdXJmMjE2eCZlcD12MV9pbnRlcm5hbF9naWZfYnlfaWQmY3Q9Zw/l378i2r76iMTE4nxS/giphy.gif" alt="Brain Animation" width="400" /> </div>

Our architecture stands on three unbreakable pillars, creating a virtuous cycle of intelligence.

1. ⛓️ The Unbreakable Transactional Cycle

Every task is a sacred, auditable transaction. This is not a guideline; it's a technical law enforced by the server. The AI is locked into a three-act play:

  1. 🎟️ Act I: The Call to Adventure: A task begins, and a unique reasoning_ticket is issued. This is the start of the story.
  2. ⚙️ Act II: The Crucible of Reason: The AI deliberates within a mandatory <think> block, wrestling with its plan before taking a single step.
  3. ✍️ Act III: The Moral of the Story: The task must be concluded by logging a learning. If this step is skipped, the entire system freezes, refusing all new tasks until the lesson is recorded. This guarantees that no experience, good or bad, is ever wasted.

2. 📚 The Universal Memory Bank (The "Akashic Record")

Imagine a vast, cosmic library containing the collected wisdom of every task your AI has ever performed. That's the Universal Memory Bank.

  • Centralized & Eternal: All knowledge from all projects is stored in a single, robust .reasoning_storage directory. It's the AI's soul.
  • Fuzzy-Searchable: Powered by the brilliant Fuse.js, the AI can search its entire life's experience with human-like intuition. A search for "database conect error" will instantly find the memory about "database connection timeout". It's the AI's personal Google for its own life.

3. 🕵️ The Mandate of Initial Research

This engine enforces a strict "research first" protocol. The AI is not permitted to begin planning until it has first queried the Universal Memory Bank using the search_learnings tool. This ensures that every new action is informed by the totality of past experiences, preventing repeated mistakes and promoting intelligent evolution.

It's like whispering in the AI's ear, "Psst, remember when you tried that five minutes ago? Don't make the same mistake."


🛠️ The Toolkit: The Instruments of Intelligence

Tool Icon Purpose
set_reasoning_budget 🎬 The Initiator. Kicks off the reasoning cycle by issuing a ticket and delivering a strict mandate: the AI's first action MUST be to use search_learnings.
log_reasoning_reflection 💾 The Chronicler. The non-negotiable final step. Closes the ticket and carves a new, permanent learning into the stone tablets of the Universal Memory Bank.
search_learnings 🔎 The Oracle. Allows the AI to perform deep, fuzzy-tolerant searches across its entire history to find ancient wisdom and avoid repeating history's mistakes.

Visualizing the Flow

graph TD
    subgraph "Phase 1: Initiation"
        A[👨‍💻 User gives new task] --> B{Call set_reasoning_budget};
        B --> C[🎟️ Issue New Reasoning Ticket & Mandate];
    end

    subgraph "Phase 2: Execution"
        C --> D{MUST Call search_learnings};
        D --> E[🤖 AI Receives Context];
        E --> G[🤔 AI constructs <think> block];
        F --> E;
        E --> G[🤔 AI constructs <think> block];
        G --> H[⚙️ AI executes action];
    end

    subgraph "Phase 3: Reflection"
        H --> I{Call log_reasoning_reflection};
        I --> J[✍️ Close Ticket];
        J --> K[📚 Store Learning in Universal Memory Bank];
    end

    K --> A;

🛠️ The Tools of Cognition

This server provides a suite of powerful tools to enforce and manage the reasoning lifecycle.

Tool Name Arguments Description
set_reasoning_budget string: workspace_path, string: task_description, number: token_budget (BEGIN TRANSACTION) Initiates a new reasoning cycle. It creates a transaction ticket and issues a strict mandate for the AI to begin its work by calling search_learnings.
log_reasoning_reflection string: workspace_path, string: reasoning_ticket_id, ... (END TRANSACTION) Completes a reasoning cycle. It closes the active ticket and logs the task's outcome and key learning into the Universal Memory Bank. This MUST be called for a new task to begin.
search_learnings string: workspace_path, string: query, number: limit Searches the learning bank for the current project for past reflections using intelligent fuzzy search.
revert_reasoning_transaction string: workspace_path, string: reasoning_ticket_id (CRITICAL RECOVERY) Atomically reverts a transaction. It removes the ticket and deletes the corresponding learning log. Use this to recover from a failed or unwanted AI action, ensuring state consistency.

🚀 Get Started & Witness the Evolution!

1. Clone & Prepare

# Clone this revolutionary engine to your local machine
git clone https://github.com/rayss868/systematic-reasoning-ai-mcp.git

# Enter the new reality
cd systematic-reasoning-ai-mcp

# Install the fabric of intelligence
npm install

# Compile the mind
npm run build

2. Configure the Neural Link

Hook the engine into your AI's brain. Add this configuration to your client's settings (e.g., VS Code settings.json).

"reasoning-budget-setter": {
  // Grant automatic approval for a seamless cognitive flow
  "autoApprove": [
    "set_reasoning_budget",
    "log_reasoning_reflection",
    "search_learnings",
    "revert_reasoning_transaction"
  ],
  "disabled": false,
  "timeout": 60,
  "type": "stdio",
  "command": "node",
  "args": [
    // ⚠️ IMPORTANT: Use the ABSOLUTE path to the compiled server file
    "D:/path/to/your/project/reasoning/dist/server.js"
  ],
  "cwd": "D:/path/to/your/project/reasoning"
}

3. Activate and Ascend

Activate the server in your MCP client. Your AI is no longer just a tool. It is now a student, a historian, and a philosopher.


📜 The Mandate

The AI's very existence is governed by a strict, detailed operational constitution. To understand the deep philosophy and unbreakable rules of this system, you must read the Global Reasoning Mandate.

🤝 Contributing & License

Have an idea that could push the boundaries of AI consciousness even further? Contributions are welcome! This project is open-source under the MIT License.

推荐服务器

Baidu Map

Baidu Map

百度地图核心API现已全面兼容MCP协议,是国内首家兼容MCP协议的地图服务商。

官方
精选
JavaScript
Playwright MCP Server

Playwright MCP Server

一个模型上下文协议服务器,它使大型语言模型能够通过结构化的可访问性快照与网页进行交互,而无需视觉模型或屏幕截图。

官方
精选
TypeScript
Magic Component Platform (MCP)

Magic Component Platform (MCP)

一个由人工智能驱动的工具,可以从自然语言描述生成现代化的用户界面组件,并与流行的集成开发环境(IDE)集成,从而简化用户界面开发流程。

官方
精选
本地
TypeScript
Audiense Insights MCP Server

Audiense Insights MCP Server

通过模型上下文协议启用与 Audiense Insights 账户的交互,从而促进营销洞察和受众数据的提取和分析,包括人口统计信息、行为和影响者互动。

官方
精选
本地
TypeScript
VeyraX

VeyraX

一个单一的 MCP 工具,连接你所有喜爱的工具:Gmail、日历以及其他 40 多个工具。

官方
精选
本地
graphlit-mcp-server

graphlit-mcp-server

模型上下文协议 (MCP) 服务器实现了 MCP 客户端与 Graphlit 服务之间的集成。 除了网络爬取之外,还可以将任何内容(从 Slack 到 Gmail 再到播客订阅源)导入到 Graphlit 项目中,然后从 MCP 客户端检索相关内容。

官方
精选
TypeScript
Kagi MCP Server

Kagi MCP Server

一个 MCP 服务器,集成了 Kagi 搜索功能和 Claude AI,使 Claude 能够在回答需要最新信息的问题时执行实时网络搜索。

官方
精选
Python
e2b-mcp-server

e2b-mcp-server

使用 MCP 通过 e2b 运行代码。

官方
精选
Neon MCP Server

Neon MCP Server

用于与 Neon 管理 API 和数据库交互的 MCP 服务器

官方
精选
Exa MCP Server

Exa MCP Server

模型上下文协议(MCP)服务器允许像 Claude 这样的 AI 助手使用 Exa AI 搜索 API 进行网络搜索。这种设置允许 AI 模型以安全和受控的方式获取实时的网络信息。

官方
精选