Skip to main content

Documentation Index

Fetch the complete documentation index at: https://docs.xano.com/llms.txt

Use this file to discover all available pages before exploring further.

This page covers using the Xano Developer MCP inside VS Code. The primary path is GitHub Copilot, since it’s the AI assistant built into VS Code itself. Other AI CLI agents (Claude Code, Codex, Gemini CLI) work just as well from VS Code’s integrated terminal — quick-setup instructions for each are at the bottom of the page.

​
Prerequisites

​
Installation

​
1. Install the MCP Server

Method 1: Click to install Install in VS Code Method 2: Manual install
  1. Open the Command Palette by pressing ⌘ / Ctrl + P and type MCP. Click MCP: Add Server
  2. Choose Command (stdio) and provide the command npx -y @xano/developer-mcp
  3. Give the server a name, such as Xano Developer MCP
  4. Choose whether to enable the MCP in the open workspace only, or for the entire profile

​
2. Install the XanoScript Language Server extension

The Language Server extension is required for the best AI-assisted XanoScript experience — it provides code completion, syntax highlighting, and inline diagnostics that every AI tool on this page (Copilot, Claude Code, Codex, Gemini CLI) reads from. Install it from the VS Code Marketplace. XanoScript Language Server

​
3. Verify Setup

In Copilot Chat, ask:
“What version of the Xano Developer MCP is installed?”
Copilot should call mcp_version and return the current version number. vs-code-copilot-20260508-114032

​
Using Copilot

The Developer MCP lets Copilot write and validate XanoScript using your local Xano workspace as context. Before asking Copilot to make changes, pull a workspace locally with the Xano CLI. If you haven’t installed and authenticated with the CLI yet, do that now before continuing. Then select or create a workspace and run: 
xano workspace pull
Once your workspace is available locally, you can ask Copilot to help with XanoScript tasks, from building a complete workspace: 
Create a small backend for advisor client intake. It should include a `clients` table with fields for name, email, phone, risk tolerance, investment objective, and advisor notes. Add API endpoints to create a client, list clients, get a single client by ID, and update client details. Include validation for required fields and valid risk tolerance values.
You can also ask it to expand or iterate on existing backends. 
Add a follow-up task feature to the advisor client intake backend. Create a `client_tasks` table linked to clients, with fields for task title, due date, status, and notes. Add API endpoints to create a task for a client, list all tasks for a client, and mark a task as complete.
Check out the full example backend on Github. Once you’re ready, push the changes to your sandbox so you can review them in the browser before promoting to your live workspace: xano sandbox review opens your sandbox in the browser — verify everything looks right, then promote the changes to production from there.

​
Using Claude Code, Gemini, Codex, or another CLI

Prefer a CLI agent over Copilot? Each of these runs in VS Code’s integrated terminal (View → Terminal). Setup is identical to the standalone install — the sections below cover the quick install plus VS Code-specific quirks. Full guides for each agent are linked at the end of every section.
Install the XanoScript Language Server extension (Step 2 above) before continuing. Every CLI agent on this page depends on it for inline syntax help and diagnostics — don’t skip it.

​
https://mintcdn.com/xano-997cb9ee/aZQYcxhIvSDTNEim/images/icons/claude-ai-icon.svg?fit=max&auto=format&n=aZQYcxhIvSDTNEim&q=85&s=b9de3a97860e4296c6d48664852a8e16 Claude Code

In the integrated terminal: Add --scope user to make the server available across every project rather than just the open workspace. → Full guide: Claude Code

​
https://mintcdn.com/xano-997cb9ee/aZQYcxhIvSDTNEim/images/icons/OpenAI_light.svg?fit=max&auto=format&n=aZQYcxhIvSDTNEim&q=85&s=8c4486c571ab4c35fe7843ed31718c04https://mintcdn.com/xano-997cb9ee/aZQYcxhIvSDTNEim/images/icons/OpenAI_dark.svg?fit=max&auto=format&n=aZQYcxhIvSDTNEim&q=85&s=21a3306ede3f58c2015e3451bc111529 Codex

Install Codex if you don’t already have it, then add the MCP — both run in the integrated terminal: → Full guide: Codex

​
https://mintcdn.com/xano-997cb9ee/iFWc-RFvMkYTid3P/images/icons/gemini.svg?fit=max&auto=format&n=iFWc-RFvMkYTid3P&q=85&s=b1868b87c460669c476c1d136ef95ebe Gemini CLI

Edit ~/.gemini/settings.json (user-wide) or .gemini/settings.json at the workspace root (project-only) and add: 
{
  "mcpServers": {
    "xano": {
      "command": "npx",
      "args": ["-y", "@xano/developer-mcp"]
    }
  }
}
Then start gemini from the integrated terminal. → Full guide: Gemini CLI

​
FAQ & Troubleshooting

1

The AI is writing invalid XanoScript, or you keep having to validate manually.

Make sure you installed the XanoScript Language Server extension. It applies to every AI tool on this page.
2

The AI doesn't seem to know about the Xano tools.

Confirm the xano MCP server is registered and connected:
  • Copilot: check the MCP servers list in VS Code’s settings.
  • Claude Code: run /mcp inside a session, or claude mcp list in the terminal.
  • Codex: run codex mcp list.
  • Gemini CLI: run /mcp inside a session.
If the server isn’t registered, re-run the install step. If it is registered but disconnected, restart the session — newly added servers don’t take effect until a fresh start.
3

I asked the AI to write a fresh backend, but it did it in something other than XanoScript.

  1. Did you install the Language Server extension?
  2. Did you authenticate with the CLI and pull a workspace (even a new, blank workspace)?
4

I want this MCP available across every project, not just one workspace.

Each tool scopes differently:
  • Copilot: during MCP install, choose the profile-wide option instead of workspace-only.
  • Claude Code: add --scope user to the install command.
  • Codex: the codex mcp add command registers the server user-wide by default.
  • Gemini CLI: put the mcpServers block in ~/.gemini/settings.json rather than .gemini/settings.json at the project root.

​
What’s Next

Tools Reference

Explore all 6 tools available in the Developer MCP

Resources & Library

Access MCP resources and use the package as an npm library