The agents project offers a unified configuration management solution designed to streamline multi-LLM development. With this tool, developers can simplify their interactions with a variety of AI platforms, including Codex, Claude, Cursor, and Gemini, through a single source of truth that enhances organization and reduces complexity.

Key Features

• Centralized Configuration: Utilize a single set of config files located in the .agents/ directory to manage all Multi-Channel Protocol (MCP) servers, skills, and instructions across different tools effortlessly.
• Automatic Synchronization: The CLI tools automatically sync MCP servers and skills, ensuring that changes made in one place are reflected across all integrated platforms.

Project Structure

.agents/
  ├── agents.json      → All your MCP servers
  ├── local.json       → Your secrets (gitignored)
  └── skills/          → Reusable team workflows

Supported Tools

Command Overview

How It Works

1. Edit the .agents/agents.json file or utilize the CLI for modifications.
2. The agents tool reads the configuration and automatically generates the specific config files needed for each integrated tool.
3. This allows developers to focus on coding without the burden of managing multiple configuration files.

Getting Started

Simply install the npm package, initialize the agents setup in your project directory, and run the synchronization command to keep everything up-to-date. This ensures that all configurations are consistently aligned across various platforms, minimizing the risks of errors due to outdated or mismatched configurations.

Community & Support

If issues arise or suggestions for improvements come to mind, users are encouraged to engage with the community by reporting bugs or participating in discussions to enhance the project further.

This repository is constructed to assist AI developers in navigating the complexities of multi-LLM environments efficiently, providing them with tools necessary for collaborative and successful developments.