Varity MCP Server

Varity Team Core Contributors Updated March 2026

The Varity MCP server (@varity-labs/mcp) gives AI editors 7 tools for the full build → deploy → monetize workflow. It runs locally via stdio — no API keys, no accounts.

Tip

The MCP server works with any editor that supports MCP: Cursor, Claude Code, VS Code (Copilot), Windsurf, and more. Setup takes under a minute.

Prerequisites

• Node.js 18+ — runs the MCP server
• Python 3.10+ — needed for varitykit CLI (used by deploy tools)
• varitykit — install with pip install varitykit

Setup

Cursor

Add to .cursor/mcp.json in your project root:

{
  "mcpServers": {
    "varity": {
      "command": "npx",
      "args": ["-y", "@varity-labs/mcp"]
    }
  }
}

Restart Cursor. The Varity tools appear in the MCP tool list.

Claude Code

Run once from any directory:

claude mcp add varity -- npx -y @varity-labs/mcp

Claude Code picks up the server automatically on the next conversation.

VS Code (Copilot)

Add to .vscode/mcp.json:

{
  "mcpServers": {
    "varity": {
      "command": "npx",
      "args": ["-y", "@varity-labs/mcp"]
    }
  }
}

Windsurf

Add to ~/.codeium/windsurf/mcp_config.json:

{
  "mcpServers": {
    "varity": {
      "command": "npx",
      "args": ["-y", "@varity-labs/mcp"]
    }
  }
}

Tools Reference

The server exposes 7 tools. Read-only tools run safely with no side effects. Destructive tools create files or deploy infrastructure.

varity_search_docs — Search documentation

Search Varity docs for guides, API references, and tutorials.

Parameter	Type	Required	Default	Description
query	string	Yes	—	Search query (e.g., “database setup”, “how to deploy”)
maxResults	number	No	3	Maximum results to return

Annotations: readOnlyHint: true

Example prompt: “Search the Varity docs for how to set up authentication”

---

varity_cost_calculator — Estimate costs

Compare Varity hosting costs against AWS, Vercel, and other platforms.

Parameter	Type	Required	Default	Description
users	number	Yes	—	Estimated monthly active users
storage_gb	number	No	10	Storage needed in GB
has_database	boolean	No	true	Whether the app uses a database
has_auth	boolean	No	true	Whether the app uses authentication

Annotations: readOnlyHint: true

Example prompt: “How much would it cost to host a SaaS app with 500 users on Varity?”

---

varity_init — Create a new app

Scaffold a production-ready app with auth, database, and payments.

Parameter	Type	Required	Default	Description
name	string	Yes	—	Project name (lowercase, hyphens allowed)
template	"saas-starter"	No	"saas-starter"	Template to use
path	string	No	Current directory	Directory to create the project in

Example prompt: “Create a new Varity app called invoice-tracker”

---

varity_deploy — Deploy to production

Build and deploy the current project. Auto-detects framework (Next.js, React, Vue), builds it, and returns a live URL.

Parameter	Type	Required	Default	Description
path	string	No	Workspace root	Path to the project directory
submit_to_store	boolean	No	false	Also submit to the Varity App Store (90% revenue to you)

Annotations: destructiveHint: true — deploys real infrastructure

Example prompt: “Deploy this project to Varity”

---

varity_deploy_status — Check deployments

List all deployments or get status of a specific one. Shows URL, status, framework, size, and creation time.

Parameter	Type	Required	Default	Description
deployment_id	string	No	—	Specific deployment ID (omit to list all)

Annotations: readOnlyHint: true

Example prompt: “Show me the status of my deployments”

---

varity_deploy_logs — Get build logs

Get build and deployment logs for debugging failed or recent deployments.

Parameter	Type	Required	Default	Description
deployment_id	string	Yes	—	The deployment ID to get logs for
limit	number	No	100	Maximum log lines to return

Annotations: readOnlyHint: true

Example prompt: “Show me the build logs for deployment abc123”

---

varity_submit_to_store — Publish to App Store

Submit a deployed app to the Varity App Store marketplace. Revenue split: 90% to you, 10% to Varity.

Parameter	Type	Required	Default	Description
deployment_id	string	Yes	—	Deployment ID from a previous varity_deploy
name	string	Yes	—	Display name on the store
description	string	Yes	—	Short description shown to users
price	number	Yes	—	Monthly price in USD (use 0 for free apps)

Annotations: destructiveHint: true — publishes to a public store

Example prompt: “Submit my last deployment to the Varity App Store at $9.99/month”

---

Response Format

All tools return JSON with a consistent structure:

// Success
{
  success: true,
  data: { /* tool-specific data */ },
  message: "Human-readable summary"
}

// Error
{
  success: false,
  error: {
    code: "ERROR_CODE",
    message: "What went wrong",
    suggestion?: "How to fix it"
  }
}

Typical Workflow

Here’s the full build-to-monetize flow using the MCP tools:

1. “Create a Varity app called my-saas” → varity_init
2. “What will this cost for 200 users?” → varity_cost_calculator
3. (Build your features)
4. “Deploy this” → varity_deploy
5. “Did the deploy succeed?” → varity_deploy_status
6. “Show me the build logs” → varity_deploy_logs
7. “Submit to the app store at $14.99” → varity_submit_to_store

Troubleshooting

“Cannot find module @varity-labs/mcp” Make sure Node.js 18+ is installed. Run node --version to check.

Deploy tools fail with “varitykit not found” Install the CLI: pip install varitykit. Run varitykit doctor to verify.

Server not showing in Cursor Restart Cursor after adding the config. Check that .cursor/mcp.json is valid JSON.

Server not showing in Claude Code Run claude mcp list to verify the server is registered. Re-add with claude mcp add varity -- npx -y @varity-labs/mcp.

Using with ChatGPT and Browser-Based LLMs

The MCP server currently supports stdio transport only, which works with local editors like Cursor, Claude Code, VS Code, and Windsurf.

For browser-based LLMs like ChatGPT or Gemini, use the llms.txt files instead:

• Summary context: https://docs.varity.so/llms.txt
• Full context: https://docs.varity.so/llms-full.txt

Paste the URL or its contents into your conversation to give the LLM full Varity context. The docs site also includes “Ask ChatGPT” links on each page that pre-load the relevant context.

Support

• Documentation
• Discord — ask in #general
• GitHub Issues