Agentsync by Dallay is a free CLI tool designed for the mcp platform, focusing on synchronizing AI agent configurations and managing MCP server definitions. Implemented in Rust with a TypeScript wrapper, it distributes via npm as "@dallay/agentsync", pnpm, Yarn, Bun, direct binaries, and source builds. This program is a client-side configuration manager, not an MCP server, offering developers a unified approach to managing diverse AI coding assistant settings and MCP server definitions across multiple environments.

What does Agentsync connect to?

Agentsync connects to various AI coding assistants, including Claude Code, GitHub Copilot, OpenAI Codex CLI, Gemini CLI, Cursor, VS Code, and OpenCode. It also supports over 32 other configurable AI agents. This CLI tool specializes in synchronizing configurations for these AI agents and managing MCP server definitions across different AI coding environments. It acts as a client-side configuration manager, handling settings for servers that utilize protocols like stdio or HTTP, rather than being a server itself.

Core features and tools

Agentsync exposes several callable tools for configuration management. Use agentsync generate to print a detailed AI prompt for project-specific configurations. The agentsync sync command reads source files and transforms them for synchronization. To add a tool, use agentsync enable <tool>, which also scaffolds its settings. Define MCP servers with agentsync add mcp <server>. Initialize configurations using agentsync init. Apply synchronized settings with agentsync apply. Use agentsync dry-run to preview changes, and agentsync check for CI-friendly validation. agentsync help displays information.

Who is it for?

Agentsync is intended for developers and teams who regularly use multiple AI coding assistants and need a centralized method for managing their configurations. It helps maintain consistency across different AI environments and simplifies the setup of MCP server definitions. The tool is ideal for those seeking to propagate AI agent instructions and manage reusable skills efficiently. Skip this if you primarily use a single AI coding assistant without needing to synchronize configurations across multiple platforms.

How to get started

No credentials are required for Agentsync itself, as it operates as a local configuration manager.

Use cases

• Maintaining a single source of truth for AI agent instructions, such as in an "AGENTS.md" file, and propagating these to various AI coding assistants.
• Defining MCP servers once in a central configuration file like "agentsync.toml" or ".ai/src/mcp.json" and automatically generating corresponding MCP configuration files for all supported AI agents.
• Installing and managing reusable AI agent skills from a curated catalog or local archives, making them available to compatible AI assistants.

Agentsync provides a unified command-line interface for managing AI agent and MCP server configurations across development environments. It simplifies the setup and synchronization of settings for numerous AI coding assistants and centralizes MCP server definitions. This tool helps developers maintain consistent AI agent behavior and skill availability.