MCP Config Generator
Setting up a Model Context Protocol server usually means hand-writing a JSON block with a server name, a launch command, arguments and environment variables, and one misplaced comma stops the client from loading it. This generator assembles that block for you in the browser, with one argument per line, environment keys turned into safe placeholders and a remote-URL option for HTTP or SSE servers. It never runs the command or contacts a server, and client-specific differences are called out so you can adapt the result.
Your workspace
Runs locally in your browserClient-specific MCP formats may differ. This page only writes JSON text: it does not run commands or contact servers. Use placeholders such as YOUR_API_KEY for secrets.
Result
Your input is processed locally in your browser and is not sent to Flutters servers. Inputs are not saved by this tool.
How to use this tool
Enter a server name, choose how the client connects, fill the command and arguments or the URL, add environment keys, then choose Generate config and copy the JSON.
Example
Input
Name: my-server
Command: npx
Arguments: -y
@scope/example-server
Environment: API_KEY=Output
{
"mcpServers": {
"my-server": {
"command": "npx",
"args": ["-y", "@scope/example-server"],
"env": { "API_KEY": "YOUR_API_KEY" }
}
}
}What does this tool do?
Model Context Protocol clients commonly read a JSON file with an mcpServers object that names each server and says how to start or reach it. This generator builds that structure from a server name, a command with arguments, optional environment variables, or a remote URL, and fills empty secrets with placeholders such as YOUR_API_KEY.
Common mistakes and limitations
Client formats differ in file location, supported fields and how remote servers are declared, so check your client's documentation and adapt the output. This tool only writes JSON text. It does not run the command, contact the URL or validate that the server exists. Replace every placeholder yourself and never commit real keys.
What the configuration describes
The Model Context Protocol lets an AI client connect to tools and data sources exposed by a server. In the common JSON layout, each server has a name under mcpServers and either a command with arguments that the client launches, or a URL for a remote server. Environment variables supply settings such as an API key without hard-coding it into the command line.
{
"mcpServers": {
"my-server": { "command": "npx", "args": ["-y", "package-name"] }
}
}Placeholders, secrets and trust
Empty environment values become placeholders like YOUR_API_KEY, and the page warns when a value looks real. Keep secrets out of files that are committed or shared. A configured command runs on your machine with your permissions when a client starts it, so only add servers you trust and read what an argument or package actually does.
Related: Prompt Template Builder
Client differences and validation
File locations, supported fields and the way remote servers are declared vary between clients and change over time, so treat the output as a starting point and compare it with your client's documentation. Format the final file and, if you maintain many configurations, describe them with a schema.
Related: JSON Formatter, JSON to JSON Schema
Frequently asked questions
Does it run my command?
No. The command is only written into JSON text. Nothing is executed or contacted.
Will it create real credentials?
No. Empty environment values become placeholders like YOUR_API_KEY, and the page warns if a value looks real.
Is this format valid for every MCP client?
Not necessarily. The mcpServers layout is common, but fields and file locations vary by client, so confirm with its documentation.
Related tools
JSON Formatter & Validator
Format and validate JSON with readable indentation.
Open tool AI & LLM ToolsJSON to JSON Schema Generator
Generate a draft JSON Schema from a sample JSON document, with nested objects, arrays and an optional required-fields toggle.
Open tool AI & LLM ToolsPrompt Template Builder
Write a prompt with {{variables}}, fill the detected fields or paste JSON values, and copy the finished prompt. Plain replacement only.
Open tool