How I Solved AI Tool Fragmentation: A Universal Sync Layer for Claude, Gemini, and Cursor

If you are using AI coding assistants today, chances are you aren't just using one. You might be using Claude Code in the terminal for rapid CLI actions, Gemini CLI for long-context search, and Cursor or Copilot as your main IDE.

But there is a major friction point: none of these tools talk to each other.

• If you configure a Model Context Protocol (MCP) server in Claude, Cursor knows nothing about it.
• If you write a detailed custom system prompt (agent instructions) for Gemini CLI, Claude starts as a blank slate.
• You end up manually copy-pasting markdown files and duplicating JSON configs across five different directories.

To solve this, I built Wasla (Arabic for connection) — a universal synchronization layer for AI coding tools.

---

What is Wasla?

Wasla is an open-source CLI and daemon that monitors and syncs your agent instructions and MCP servers across Claude Code, Gemini CLI, Cursor, and more.

It runs entirely locally, takes seconds to set up, and ensures your assets are shared across every tool you use.

# Set up Gemini in your workspace
npx @untitled-devs/wasla setup gemini --scope workspace

---

How it Works Under the Hood

Instead of copying entire folders (which leads to file duplication and conflicts), Wasla uses a "Latest is Greatest" strategy and lightweight reference stubs:

1. Asset Discovery: Wasla scans your configured tool directories (like ~/.claude/ and ~/.gemini/).
2. Reference Stubs: Instead of copying whole files, it writes a lightweight reference pointer. When Claude loads the stub, it points directly to the original file created in Gemini.
3. Automatic Mirroring: Whichever tool you edited most recently becomes the source of truth for the next sync.

Here is a quick look at the directory mapping:

~/.gemini/agents/planner.md    <-- Original file
~/.claude/agents/planner.md    <-- Stub pointing to Gemini original
~/.openclaw/agents/planner.md  <-- Stub pointing to Gemini original

---

Key Features

1. Dual-Scope Syncing

You can choose how you sync your configurations:

• --scope workspace: Keep prompts and configs local to your active project/git repository.
• --scope user: Share your custom agents globally across your entire machine.

2. Daemon Watch Mode

Run wasla watch. It runs quietly in the background, listening for file saves, and updates every tool's directory instantly.

3. Interactive Web Dashboard

Run wasla visualizer. It spins up a local React dashboard showing exactly which agents, instructions, and MCP servers are connected to which tools.

---

Getting Started

You can install it globally or run it directly using npx:

# Keep user-level tools in sync in the background
npx @untitled-devs/wasla watch --scope user

The project is fully open-source (MIT). I'd love to hear your thoughts, feedback on the architecture, or ideas for new adapters!

Check out the repo here: github.com/The-Untitled-Org/wasla