Free tool, no signup
MCP config generator
This free MCP config generator writes the server config each AI client expects. Describe your server once, either a command to run or a URL to call, and copy the exact JSON or TOML for Claude Desktop, Claude Code, Cursor, VS Code and Codex, along with where to paste it. Any keys you type stay in your browser.
MCP config generator
Start from a known server
Letters, numbers, - and _. This is the name the client shows.
KEY=value. Stays in your browser.
Config for filesystem is ready for 5 clients. Open the one you use, copy it and paste it where it says.
Claude Desktop
- macOS: ~/Library/Application Support/Claude/claude_desktop_config.json
- Windows: %APPDATA%\Claude\claude_desktop_config.json
{
"mcpServers": {
"filesystem": {
"command": "npx",
"args": [
"-y",
"@modelcontextprotocol/server-filesystem",
"/Users/you/Documents"
]
}
}
}Quit and reopen Claude Desktop after saving.
Claude Code
- Run the command in your project, or commit .mcp.json at the project root to share it
claude mcp add --transport stdio filesystem -- npx -y @modelcontextprotocol/server-filesystem /Users/you/Documents{
"mcpServers": {
"filesystem": {
"command": "npx",
"args": [
"-y",
"@modelcontextprotocol/server-filesystem",
"/Users/you/Documents"
]
}
}
}Cursor
- This project: .cursor/mcp.json
- Every project: ~/.cursor/mcp.json
{
"mcpServers": {
"filesystem": {
"type": "stdio",
"command": "npx",
"args": [
"-y",
"@modelcontextprotocol/server-filesystem",
"/Users/you/Documents"
]
}
}
}VS Code
- This workspace: .vscode/mcp.json
{
"servers": {
"filesystem": {
"type": "stdio",
"command": "npx",
"args": [
"-y",
"@modelcontextprotocol/server-filesystem",
"/Users/you/Documents"
]
}
}
}Codex
- Every project: ~/.codex/config.toml
- A trusted project: .codex/config.toml
[mcp_servers.filesystem]
command = "npx"
args = ["-y", "@modelcontextprotocol/server-filesystem", "/Users/you/Documents"]These clients only use your tools while your computer is on and the app is open. Construct gives an AI agent its own cloud computer, connects any remote MCP server by its HTTPS URL, and runs scheduled jobs with your laptop closed.
Connect your server to ConstructMCP config file locations
Where each client reads its MCP servers, and the key it expects at the top of the file.
| Client | Config file | Top-level key | Remote servers |
|---|---|---|---|
| Claude Desktop | Mac: ~/Library/Application Support/Claude/claude_desktop_config.json. Windows: %APPDATA%\Claude\claude_desktop_config.json | mcpServers | Settings, Connectors, Add custom connector |
| Claude Code | .mcp.json at the project root, or claude mcp add | mcpServers | "type": "http" with "url" |
| Cursor | .cursor/mcp.json in a project, or ~/.cursor/mcp.json for all projects | mcpServers | "url" with optional "headers" |
| VS Code | .vscode/mcp.json in the workspace | servers | "type": "http" with "url" |
| Codex | ~/.codex/config.toml, or .codex/config.toml in a trusted project | [mcp_servers.<name>] | url with optional http_headers |
Frequently asked questions
Where is the Claude Desktop config file?
On a Mac it's ~/Library/Application Support/Claude/claude_desktop_config.json, and on Windows it's %APPDATA%\Claude\claude_desktop_config.json. The quickest way to open it is Claude's Settings, then the Developer tab, then Edit Config. If the file doesn't exist yet, Edit Config creates it. Quit and reopen Claude after you save.
How do I add a remote MCP server to Claude Desktop?
Remote servers don't go in claude_desktop_config.json. Open Settings, then Connectors, click Add and choose Add custom connector, paste the server's URL and click Add. If the server needs you to sign in, Claude walks you through it.
What's the difference between a local and a remote MCP server?
A local server is a program the client starts on your computer, usually with npx, uvx or docker, and it talks to the client directly. A remote server runs somewhere else and you reach it by URL over HTTP. Local servers can read files on your machine, while remote ones keep working when your computer is off and a whole team can share them.
How do I convert mcp.json to Codex's config.toml?
Each server in the mcpServers object becomes a [mcp_servers.<name>] table in ~/.codex/config.toml, with command and args as keys and environment variables in a [mcp_servers.<name>.env] table. Remote servers use url and http_headers. Describe the server once on this page and the Codex tab gives you the TOML.
Should I commit .mcp.json or add it to .gitignore?
Commit it when it holds no secrets, since that's how Claude Code shares project servers with your team. Keys sit in these files as plain text, so keep them out of anything you commit. Files in your home folder, like ~/.cursor/mcp.json or ~/.codex/config.toml, stay on your machine.
My MCP server doesn't show up. What should I check?
Quit the client completely and open it again, then check the file is valid JSON or TOML. Run the same command yourself in a terminal to see any errors it prints. File paths in arguments should be absolute, not relative.
Is it safe to type API keys into this page?
Nothing you type is sent anywhere. The config is built in your browser, and this part of the page is excluded from our analytics and session recordings. We only count that a snippet was copied, never what was in it.
More free tools
- EmailGmail filter generatorBuild Gmail filters for multiple senders, domains and subjects, test each one in Gmail search, then import them all with one XML file. Free, no signup.
- AutomationZapier cost calculatorHow much will Zapier cost you? Enter your Zaps, steps and runs to see your monthly tasks, the cheapest plan and your cost per task, at current Zapier prices.
Related reading
- What is an AI agent with its own computer?
An AI agent with its own computer keeps working in a cloud workspace after you close your laptop. How it works, the risks, and who offers one, from $9 a month.
- All our Agents get computers, we pay for almost none
Construct is an AI employee with a real Linux computer per user. Don't get excited - we are not paying for all of them. How we run that on Cloudflare Durable Objects, Sandboxes, and R2.
- Construct vs coding-focused agents
Compare Construct with coding agents. See when a generalist AI employee for email, research, calendar, CRM, and code is the better fit.