What is MCP (Model Context Protocol)?
The Model Context Protocol (MCP) is an open standard that defines how AI applications communicate with external tools and services. Think of it as a "USB-C port for AI" — instead of every AI app building custom integrations for every tool, MCP provides a single, standardized protocol.
This editor is the power-user companion to the guided MCP Config Generator. Where the generator walks you through a wizard, the editor lets you work directly with the raw mcpServers JSON, convert the same config between client formats, and validate it before dropping it into your AI app.
Edit, Validate, Convert
JSON-First Editing
The editor is built around the MCP config JSON. You can:
- Paste an existing
claude_desktop_config.json,opencode.json,.cursor/mcp.json, or VS Codesettings.json - Switch to the form view to add, remove, and edit servers without hand-writing JSON
- Watch validation errors and warnings update as you type
Multi-Server, Multi-Client
A single config can contain many servers. The editor keeps them all in one place and converts the whole document for whichever client you target:
| Client | Config File | Root Key |
|---|---|---|
| Claude Desktop | claude_desktop_config.json |
mcpServers |
| OpenCode | opencode.json |
mcpServers |
| Cursor | .cursor/mcp.json |
mcpServers |
| VS Code | settings.json |
mcp.servers |
| Other | mcp-config.json |
mcpServers |
Switching the target client rewrites the output (including the VS Code mcp.servers key) without touching your source config.
Validation & Health Checks
The editor reuses the same validation engine as the generator:
- JSON syntax checking
- Per-server checks (command for stdio, url for sse/http)
- Empty-secret warnings for env vars that look sensitive
- Template-matched health checks (missing tokens, Node/uv requirements, OS notes)
Environment Variables & .env.example
Secrets never belong in committed config. The editor collects every env key across all servers and generates a .env.example you can copy or download, so you can keep real values out of version control.
Shareable Configs
Copy a share link to send a config to a teammate. The entire config is compressed into the URL hash with lz-string — nothing is uploaded to any server.
Security Best Practices
- Use dedicated tokens: Create API tokens with minimal required scopes instead of using personal access tokens
- Read-only when possible: For databases, create a read-only user if you only need query access
- Restrict filesystem access: Only grant access to directories the AI actually needs to read
- Review permissions: Check the health checks before adding any MCP server
- Revoke unused access: Remove MCP server configs you no longer use
- Keep tokens secret: MCP configs contain tokens — use
.env.exampleand don't commit real values
Resources
- MCP Specification — Official documentation and specification
- MCP GitHub — Open source SDKs and examples
- Claude Desktop MCP Setup — Quickstart guide