DEV Community

0xLivian
0xLivian

Posted on

Introducing Solana Instruction MCP — A Game-Changing Tool for Solana Developers

#ai #solana #mcp

I had the honor of meeting the author of this project during the Chengdu HackHouse, and after testing the Solana Instruction MCP, I was genuinely blown away.
This project is so impactful that it would be a real loss for the Solana and AI developer community if it goes unnoticed.

So, I decided to write this article to introduce how to install and use this powerful MCP.

⚙️ Note

This guide currently supports only Linux and macOS users.
Instructions for IDE integrations (e.g., VS Code) will be added later.

The project already has an official website, which is currently simple but functional — the author mentioned that it will be updated and beautified later.

Official Website: https://solmcp.daog1.workers.dev/dashboard
Author: Xiaodao (Community ID)

🔑 Step 1: Log in and Get Your API Key

Go to the official website and click Log In with GitHub.
Once logged in, you’ll see a dashboard page like this:

After login, you’ll be able to see your API Key:

💻 Step 2: Set Up with Your AI CLI (Example: Opencode CLI)

Depending on which AI CLI tool you use, check its documentation on how to connect an MCP server.

Here we’ll take Opencode CLI as an example.

First, visit the official Opencode CLI website and install it following their instructions.
After installation, you can verify it by running a version check or chatting with Opencode via terminal.

🔧 Step 3: Configure the Solana MCP

To install and configure an MCP, refer to the official Opencode documentation here:
👉 https://opencode.ai/docs/mcp-servers/#remote

Look for the section titled “Using opencode.json” — this explains how to add MCP servers manually.

You’ll need to create (or edit) an opencode.json file inside either:

  • your project’s .opencode directory, or
  • your global config directory: ~/.opencode

Global installation is recommended. Run the following commands:

cd ~/.opencode
touch opencode.json
Enter fullscreen mode Exit fullscreen mode

Then insert the following content:

{
  "$schema": "https://opencode.ai/config.json",
  "mcp": {
    "SolanaMCP": {
      "type": "http",
      "url": "https://solmcp.daog1.workers.dev/mcp?apiKey=MY_API_KEY",
      "enabled": true,
      "headers": {
        "Authorization": "Bearer MY_API_KEY"
      }
    }
  }
}
Enter fullscreen mode Exit fullscreen mode

Replace MY_API_KEY with the key displayed on the MCP dashboard (after apiKey=).

Save the file, then restart Opencode by typing:

opencode
Enter fullscreen mode Exit fullscreen mode

Once restarted, you can start using the Solana MCP!
When giving instructions to your AI assistant, make sure to explicitly mention “using MCP” or similar wording so it recognizes the request correctly.

✅ Summary

That’s it! You now have Solana Instruction MCP connected to your AI CLI environment.
With it, your assistant can perform deep instruction-level analysis of Solana transactions, decode parameters, trace CPIs, and much more — all powered by an expanding IDL library and AI-enhanced understanding.

If you’re building tools or doing audits in the Solana ecosystem, this is a must-try.

Top comments (0)