A terminal session wiring an AI coding agent into external tools

Claude Code MCP: Add Servers From the Command Line

Claude Code adds an MCP server in a single command, and that is genuinely the whole setup. No JSON file to hand-edit, no settings screen, no restart ritual. It is the fastest of any MCP client, and it is a completely different mechanism from the custom connectors you add on claude.ai.

That last point causes real confusion, so it is worth settling first.

Claude Code MCP is not claude.ai MCP

They share a protocol and nothing else. A custom connector added on claude.ai lives in your Anthropic account and is brokered from Anthropic’s cloud. A server added in Claude Code lives in your local configuration and runs from your machine.

Adding a server in one does not add it to the other. If you configured something in the browser and cannot see it in the terminal, that is why, and no amount of re-authenticating will bridge them.

Adding a server

For a remote server over HTTP:

claude mcp add --transport http carly https://carlyassistant.com/mcp/

For a local server that runs as a process, pass the command after --:

claude mcp add myserver -- npx -y @example/mcp-server

Then confirm what is loaded:

claude mcp list

Inside a session, /mcp shows the connected servers and lets you authenticate any that need OAuth. A server that appears in list but not in /mcp has almost always failed to start, and /mcp is where the error surfaces.

Scope decides who else gets the server

This is the flag worth learning early, because getting it wrong is how a server mysteriously disappears when you change directory.

  • Local is the default. The server exists for you, in the current project only.
  • Project writes to a checked-in file, so everyone working in that repository gets the same server. This is the one for a team.
  • User makes the server available to you across every project on the machine.

Pass it with --scope, for example claude mcp add --scope user .... Reach for user scope for anything you want everywhere, such as an assistant that holds your accounts, and project scope for anything specific to a codebase.

What the connection will not do

There is no such thing as an MCP trigger. The protocol's own working group says plainly that clients learn about server-side changes by polling for them, and a real event mechanism remains unshipped. Carly is where that part lives, watching thousands of apps and starting the work itself.

Claude Code separately has its own ways to run work on a schedule or on repository events, and those are worth using. They are not MCP, though, and conflating the two leads people to expect a server to wake up on its own. It will not. An MCP server sits there until the agent calls it.

Give Claude Code an assistant that runs when you are not at the keyboard

Carly is a remote MCP server holding your mail, calendars, files, and business tools, with workflows that start on real events rather than on a prompt.

  1. Add your accounts to Carly. Sign in at carlyassistant.com, open Integrations, pick a service, click Connect, and authorize. Click Connect again for the next one and keep going for as many as you want. Carly connects to thousands of apps, so everything sits behind one server.
  2. Add Carly to Claude Code. One command, at user scope so it follows you across projects:
claude mcp add --scope user --transport http carly https://carlyassistant.com/mcp/
  1. Run /mcp to authenticate, then ask. The agent can check your calendar before it proposes a meeting, read the thread that describes a bug, or file the result, without you leaving the terminal.

Free Zapier-style workflows; AI agents from $35/month. Connecting Carly costs nothing.

Frequently Asked Questions

How do I add an MCP server to Claude Code?

Run claude mcp add <name> -- <command> for a local server, or claude mcp add --transport http <name> <url> for a remote one. Check the result with claude mcp list.

Why doesn’t Claude Code see the connector I added on claude.ai?

They are separate mechanisms. Custom connectors on claude.ai are brokered through your Anthropic account; Claude Code servers live in your local configuration. Servers must be added to each.

What is the difference between local, project, and user scope?

Local is you in one project and is the default. Project is checked in so your whole team gets it. User is you across every project on the machine.

How do I see which MCP servers are connected?

claude mcp list outside a session, or the /mcp command inside one, which also handles OAuth for servers that need it.

Can an MCP server in Claude Code run on a trigger?

No. MCP has no scheduler and no event mechanism, so the server responds only when the agent calls it.


More: Claude MCP · How to install an MCP server in 7 AI clients · Cursor MCP · MCP servers: the complete list · What is MCP?

Ready to automate your busywork?

Carly schedules, researches, and briefs you—so you can focus on what matters.

See what people say

"Before Carly, I relied on a Calendly link, but the whole process felt impersonal and not very professional. Carly changed that by handling all the back-and-forth, so I'm no longer stuck in endless email threads trying to line up schedules.

Now Carly reaches out to candidates, shares my real-time availability, lets them pick a slot, then sends a Zoom link and drops it straight into my calendar. She sends reminders to both of us before each call, which has significantly reduced no-shows and last-minute confusion.

On top of scheduling, Carly acts like a full executive assistant, sending me my schedule the night before so I can prepare for each call. It reminds me of the old x.ai assistant, but Carly is noticeably smarter, faster, and better suited to my healthcare recruitment business."

Gus Ibrahim, Founder & Director, IHR