mcpfold for solo developers
If you use more than one AI client, mcpfold keeps your MCP servers in one config and folds them out to every client — so you set a server up once, not once per tool.
You run Claude Code, Cursor, and VS Code — and each keeps its MCP servers in its own file and its own shape. Adding one server everywhere means editing several config files by hand and keeping them in sync.
With mcpfold you maintain one canonical `mcp.config.jsonc` and run `mcpfold sync`; each client’s native config is written from it. Curate the tools you actually use to keep your context window lean, and keep API tokens as `${env:…}` / `${op:…}` references instead of pasting them into config files.
Everything runs locally on your machine — no account, no cloud required. Install it and import what you already have in one step.
- One config, every client. Write servers once; mcpfold renders each client’s native format.
- Curate tools. Allow/deny tools per server so only what you need loads.
- Secrets as references. Tokens stay as references, never written into a client config.
- Local-first. No account needed — the CLI is free and MIT-licensed.