One gateway for every MCP server
Running more than one AI coding tool means configuring the same MCP servers over and over. Conduit fixes that with a single local gateway.
If you use more than one AI coding tool — say Claude, Cursor, and VS Code — you’ve hit this: every client wants its own MCP configuration. You add the same servers several times, re-authenticate in each, and then drown every agent in hundreds of tool definitions it will never use.
Conduit is the fix. It’s a local-first MCP gateway: you set up and authenticate each server once, point every client at the single Conduit gateway, and every server is instantly available everywhere.
Set up once, use everywhere
Add a server and authenticate it a single time. It shows up in every connected client, with no per-app config and no restart when you add or toggle one.
Small context, not hundreds of tools
In lazy-discovery mode the gateway advertises just three meta-tools —
conduit_status, conduit_search_tools, and conduit_call_tool — instead of the
full catalog. The agent searches and calls on demand, so context stays flat no
matter how many servers you connect.
Keys stay yours
Clients only ever say “talk to Conduit.” OAuth tokens and API keys live in your OS keychain and are injected at runtime — never written into a client config file. Every tool call is recorded in an audit log.
Conduit is free and open source (MIT), for Windows, macOS, and Linux. It’s the studio’s flagship, and it’s the kind of tool I built because I needed it myself.