Nsaf MCP Server

🚀 NSAF MCP Server

This is a Model Context Protocol (MCP) server implemented for the Neural Symbolic Autonomous Framework (NSAF). It enables AI assistants to interact with the NSAF framework via the MCP protocol.

⚠️ Important Note

This repository contains both the NSAF framework code and the MCP server implementation, making it a complete package that can be deployed and used anywhere.

⚠️ Important Note

This implementation uses a simplified version of the MCP protocol and does not require the official MCP SDK. It implements the most core functions necessary for integration with NSAF features.

✨ Features

• Run NSAF evolution with customizable parameters.
• Compare different NSAF agent architectures.
• Integrate NSAF features into AI assistants.

📦 Prerequisites

• Node.js 18+ and npm
• Python 3.8+ with the NSAF framework installed

📦 Installation

1. Clone this repository:

git clone https://github.com/ariunbolor/nsaf-mcp-server.git
cd nsaf-mcp-server

2. Install dependencies:

npm install

3. Build the server:

npm run build

🛠️ Configuration

The server includes the NSAF framework code, so no additional configuration is required for basic use. The MCP server is designed to work out of the box.

💻 Usage Examples

Basic Usage

Run the server locally

npm start

Deploy to GitHub

1. Create a new GitHub repository for your MCP server:

   • Visit GitHub and create a new repository named nsaf-mcp-server.
   • Initialize it and add a README file.

2. Push the code to GitHub using the provided setup script:

# For a new repository
./setup-github-fixed.sh yourusername

# If the repository already exists and you want to overwrite the content
./setup-github-fixed.sh yourusername --force

This script performs the following actions:

• Initializes git (if necessary).
• Sets up the remote repository.
• Commits changes.
• Attempts to push to GitHub (with options to handle existing repositories).

3. Configure GitHub Actions for CI/CD (optional):
   • Create a .github/workflows directory.
   • Add workflow files for testing and building the server.

Use with an AI assistant

To use this MCP server with an AI assistant (such as Claude), follow these steps:

1. Install the server: Option 1: Install from GitHub (after pushing the code):

   npm install -g yourusername/nsaf-mcp-server
   

   Option 2: Install from a local directory:

   # Navigate to the nsaf_mcp_server directory
   cd nsaf_mcp_server
   
   # Install dependencies and build
   npm install
   npm run build
   
   # Install globally from the local directory
   npm install -g .
   

2. Add the server to the MCP settings configuration: For the Claude Desktop app (on macOS): Edit the ~/Library/Application Support/Claude/claude_desktop_config.json file:

{
  "mcpServers": {
    "nsaf": {
      "command": "nsaf-mcp-server",
      "args": [],
      "env": {},
      "disabled": false,
      "port": null
    }
  }
}

For other platforms, refer to the relevant documentation.

Advanced Usage

Run NSAF evolution (run_nsaf_evolution)

node run-nsaf-evolution.js --population-size=100 --generations=50 --mutation-rate=0.1

Parameters:

• --population-size: Population size (default: 100)
• --generations: Number of evolution generations (default: 50)
• --mutation-rate: Mutation rate (default: 0.1)

Compare NSAF agent architectures (compare_nsaf_agents)

node compare-nsaf-agents.js --agent1=agent1.json --agent2=agent2.json

Parameters:

• --agent1: Path to the first agent file
• --agent2: Path to the second agent file

📄 License

This project is licensed under the MIT License.