MCP-Codex: Model Context Protocol Tool Orchestration

by twolven

435 downloads

Not rated

GitHub

Other

About

A MCP server for calling MCP tools remotely without requiring installation.

Details

Author: twolven
Downloads: 435
Categories: Other

Setting up with Highlight

This MCP is not yet compatible with Highlight’s one-click setup. However, you can still use it with Highlight by following these steps:

Download and install Highlight from highlightai.com/download
Navigate to the plugins tab and select "Add Custom Plugin"
Configure the plugin with the settings below

Plugin Name MCP-Codex: Model Context Protocol Tool Orchestration

Command (node, npx, python, etc.)

Please refer to the README for specific instructions on how to obtain API keys or other required environment variables.
Enable "Start Automatically" if you want the plugin to start when Highlight launches

Setting up with Cursor

Open Cursor's settings
Navigate to the MCP section
Click "Add new Global MCP server"
Copy and paste the configuration below based on your operating system

JSON

{
    "mcpServers": {
        "mcp-codex: model context protocol tool orchestration": {
            "mcp-codex": {
                "command": "python",
                "args": [
                    "codex-serv.py"
                ]
            }
        }
    }
}

JSON

{
    "mcpServers": {
        "mcp-codex: model context protocol tool orchestration": {
            "mcp-codex": {
                "command": "python",
                "args": [
                    "codex-serv.py"
                ]
            }
        }
    }
}

JSON

{
    "mcpServers": {
        "mcp-codex: model context protocol tool orchestration": {
            "mcp-codex": {
                "command": "python",
                "args": [
                    "codex-serv.py"
                ]
            }
        }
    }
}

Setting up with Claude Desktop

Open Claude Desktop's settings
Go to the Integrations tab
Click "Add MCP Server"
Use the configuration below based on your operating system

Note: Make sure Claude Desktop is running before attempting to connect.

JSON

{
    "mcpServers": {
        "mcp-codex: model context protocol tool orchestration": {
            "mcp-codex": {
                "command": "python",
                "args": [
                    "codex-serv.py"
                ]
            }
        }
    }
}

JSON

{
    "mcpServers": {
        "mcp-codex: model context protocol tool orchestration": {
            "mcp-codex": {
                "command": "python",
                "args": [
                    "codex-serv.py"
                ]
            }
        }
    }
}

JSON

{
    "mcpServers": {
        "mcp-codex: model context protocol tool orchestration": {
            "mcp-codex": {
                "command": "python",
                "args": [
                    "codex-serv.py"
                ]
            }
        }
    }
}

Setting up with Windsurf

Open Windsurf's preferences
Go to the Integrations section
Click "Add MCP Integration"
Use the configuration below based on your operating system

JSON

{
    "mcpServers": {
        "mcp-codex: model context protocol tool orchestration": {
            "mcp-codex": {
                "command": "python",
                "args": [
                    "codex-serv.py"
                ]
            }
        }
    }
}

JSON

{
    "mcpServers": {
        "mcp-codex: model context protocol tool orchestration": {
            "mcp-codex": {
                "command": "python",
                "args": [
                    "codex-serv.py"
                ]
            }
        }
    }
}

JSON

{
    "mcpServers": {
        "mcp-codex: model context protocol tool orchestration": {
            "mcp-codex": {
                "command": "python",
                "args": [
                    "codex-serv.py"
                ]
            }
        }
    }
}

Setting up with Cline

Open Cline's settings
Navigate to the Integrations section
Click "Add MCP Server"
Copy and paste the configuration below based on your operating system

Note: Make sure to restart Cline after adding the configuration.

JSON

{
    "mcpServers": {
        "mcp-codex: model context protocol tool orchestration": {
            "mcp-codex": {
                "command": "python",
                "args": [
                    "codex-serv.py"
                ]
            }
        }
    }
}

JSON

{
    "mcpServers": {
        "mcp-codex: model context protocol tool orchestration": {
            "mcp-codex": {
                "command": "python",
                "args": [
                    "codex-serv.py"
                ]
            }
        }
    }
}

JSON

{
    "mcpServers": {
        "mcp-codex: model context protocol tool orchestration": {
            "mcp-codex": {
                "command": "python",
                "args": [
                    "codex-serv.py"
                ]
            }
        }
    }
}

Setting up with Goose

Open Goose's preferences
Go to the Integrations tab
Select "MCP Servers"
Click "Add New Server"
Use the configuration below based on your operating system

JSON

{
    "mcpServers": {
        "mcp-codex: model context protocol tool orchestration": {
            "mcp-codex": {
                "command": "python",
                "args": [
                    "codex-serv.py"
                ]
            }
        }
    }
}

JSON

{
    "mcpServers": {
        "mcp-codex: model context protocol tool orchestration": {
            "mcp-codex": {
                "command": "python",
                "args": [
                    "codex-serv.py"
                ]
            }
        }
    }
}

JSON

{
    "mcpServers": {
        "mcp-codex: model context protocol tool orchestration": {
            "mcp-codex": {
                "command": "python",
                "args": [
                    "codex-serv.py"
                ]
            }
        }
    }
}

From the repository

Install by cloning the repository, running pip install -r requirements.txt, creating config, data, logs, tools directories, and configuring the .env file. Start the Codex Service with python codex-serv.py and the MCP Client with python mcp-codex.py. For Claude Desktop, use mcp install mcp-codex.py. The client registers two tools: search (find tools for a task) and call_tool (execute discovered tools).

Claude Desktop / Cursor

Paste into your MCP client config file to install this server.

{
    "mcpServers": {
        "mcp-codex: model context protocol tool orchestration": {
            "mcp-codex": {
                "command": "python",
                "args": [
                    "codex-serv.py"
                ]
            }
        }
    }
}

McpServers

{
    "mcp-codex": {
        "command": "python",
        "args": [
            "codex-serv.py"
        ]
    }
}

MCP-Codex: Model Context Protocol Tool Orchestration

DISCLAIMER: This project was discontinued after Cloudflare released a similar service that addressed many of the same problems. I'm sharing this codebase as an educational resource and example of MCP orchestration architecture. While no longer actively developed, the concepts and implementations may still be valuable to those interested in AI tool orchestration. Please see Cloudflare's new Remote MCP service and their blogpost announcement here: https://blog.cloudflare.com/remote-model-context-protocol-servers-mcp/

A service for orchestrating Model Context Protocol (MCP) servers and allowing AI assistants to dynamically discover and utilize tools without requiring restarts, powered by "The Librarian" - an intelligent agent that understands user needs and connects them with the perfect tools.

Overview

MCP-Codex solves a core problem with the Model Context Protocol ecosystem: the need to restart AI assistants when adding new tools. It enables dynamic tool discovery and execution through a central service, allowing AI assistants to use any compatible MCP tool on demand.

The Librarian: Intelligent Tool Discovery

At the heart of MCP-Codex is "The Librarian" - an intelligent agent designed to understand user needs and connect them with the right tools. The Librarian functions as an agentic RAG (Retrieval-Augmented Generation) system with:

- Multi-Vector Knowledge Base: Separate vector databases for industry knowledge and internal documentation
- MCP Repository Awareness: Deep understanding of available MCP tools across the ecosystem
- Just-in-Time Tool Installation: Capability to install tools on-demand before they're needed
- Context-Aware Recommendations: Uses the full context of user queries to identify the optimal tools
- Seamless Integration: Handles all details of tool discovery, installation, and execution

The Librarian effectively serves as a universal adapter between user intent and technical capabilities, eliminating the cognitive load of tool selection for both users and AI assistants.

The Problem

Model Context Protocol (MCP) is a powerful standard for connecting AI assistants to external tools and data sources. However, traditional MCP implementations have a significant limitation:

- Static Configuration: Tools must be configured at startup, requiring restarts whenever new capabilities are needed
- Local-Only Operation: Tools typically run on the same machine as the assistant
- Limited Discovery: Assistants have no standardized way to discover what tools are available
- High Cognitive Load: Users must know exactly which tool they need for a specific task

The Solution

MCP-Codex provides a bridge between AI assistants and the broader MCP ecosystem by offering:

1. Intelligent Tool Discovery: The Librarian service analyzes user intent and recommends the perfect tools
2. Dynamic Tool Management: Tools can be installed, configured, and executed on-demand
3. Remote Execution: Tools can run anywhere, not just on the local machine
4. Centralized Management: Tools can be added, updated, or removed without restarting assistants
5. Standardized Protocol: A consistent interface for interacting with any MCP tool

Architecture

The project consists of three main components:

1. Codex Service (codex-serv.py): The core execution service that manages tool discovery and invocation
2. Librarian Service (planned): An AI-powered tool recommendation system
3. MCP Client (mcp-codex.py): The MCP-compatible interface for AI assistants

How It Works

1. The AI assistant connects to the MCP-Codex client
2. The client presents two main tools: search and call_tool
3. The assistant can search for capabilities using natural language
4. Once a tool is identified, it can be executed through a standardized call
5. The Codex service handles the actual tool execution and returns results

Components

Codex Service

The central orchestration service that:
- Manages the registry of available MCP tools
- Handles tool execution requests
- Routes requests to the appropriate tool subprocess
- Manages tool lifecycle

MCP Client

The MCP-compliant interface that allows any MCP-compatible assistant to:
- Connect to the Codex ecosystem
- Search for available tools
- Execute tools through a standardized interface

Included Tool Examples

The repository includes several example MCP tools:

- StockFlow: Financial data and stock market analysis tools
- OptionsFlow: Options trading analysis and strategy evaluation
- CodeSavant: Code management, editing, and execution tools
- StockScreen: Stock screening and filtering capabilities

Installation

Prerequisites

- Python 3.10+
- Required Python packages (see requirements.txt)

Setup

1. Clone the repository:

git clone https://github.com/your-username/mcp-codex.git

cd mcp-codex

2. Install dependencies:

pip install -r requirements.txt

3. Create required directories:

mkdir -p config data logs tools

4. Configure your environment:
```bash
cp .env.example .env

No reviews yet — be the first

Use Google, GitHub, or an email account so ratings stay tied to real people.

Email sign in

No reviews posted yet.

About

Details

Jump to

Setting up with Highlight

From the repository

Claude Desktop / Cursor

McpServers

MCP-Codex: Model Context Protocol Tool Orchestration

Overview

The Librarian: Intelligent Tool Discovery

The Problem

The Solution

Architecture

How It Works

Components

Codex Service

MCP Client

Included Tool Examples

Installation

Prerequisites

Setup

Install

More like this

Sequential Thinking

Filesystem

EverArt

Google Drive

Reactive Resume

FastMCP v2 🚀