Connect Claude Code

Claude Code connects to Appentic as an MCP server. You add the server once, sign in through your browser, and the connection then works in every project on your computer.

Before you start

You need credits on your balance. An agent can add machines, and a machine is billed from the moment it's ready, whether or not anything runs on it yet. See Adding credits.

1. Add the server

In your terminal, run:

claude mcp add --transport http --scope user appentic https://mcp.appentic.com/mcp

This only saves the server to Claude Code's settings. You sign in in the next step.

--scope user makes the server available in every project. Without it, Claude Code adds the server only to the folder you ran the command in, and a session opened elsewhere won't show it.

2. Sign in

Sign in from claude in a terminal. The VS Code and JetBrains extensions can't run the sign-in, but they use the connection once it's made.

1. Start Claude Code, or restart it if it was already open.
2. Run /mcp.
3. Choose appentic, then Authenticate.
4. Your browser opens an Appentic page that names the agent, such as Connect Claude Code?, and the account it will act as, under Connecting as. Check it, then click Connect.

Not now refuses the connection.

Which account it connects to

The agent connects to whichever Appentic account your browser is signed in to when you click Connect. That's separate from the account you use for Claude itself. Appentic never sees your Anthropic login, and the two don't have to match.

If you have more than one Appentic account, check the email under Connecting as before approving. Use a different account signs you out and brings you back to the same page so you can sign in as the one you meant. If you aren't signed in to Appentic at all, it asks you to sign in first and then brings you back.

Claude Code stores the connection and refreshes it by itself, so you aren't asked again unless you disconnect it or leave it unused for 30 days. /mcp then shows appentic as connected, and it appears in Appentic under User settings → Connected accounts.

3. Ask for something

Try a small request first, so you can see what the agent does before it spends anything:

What can you do on Appentic? Read the start guide first.

Then something real:

Create an Appentic project called stock-tracker, price a small machine
with Postgres before you add anything, and tell me what it would cost.

The agent quotes the plan and waits. Appentic refuses to add anything that costs money until the agent confirms you agreed.

See and remove connections

Go to User settings → Connected accounts. Under Coding agents, Connected agents lists each agent you approved, with when it was connected and last used. Click its disconnect button to stop it reaching your projects straight away.

A token for a script

For something that can't open a browser, such as a script, a scheduled job or an agent on a server, make a token instead.

1. Go to User settings → Connected accounts.
2. Under Coding agents, in Tokens, type what the token is for in Description and click Make a token.
3. Copy it. It isn't shown again.

Send it as a bearer token:

curl https://mcp.appentic.com/mcp \
  -H "Authorization: Bearer $APPENTIC_TOKEN" \
  -H "Content-Type: application/json" \
  -H "Accept: application/json, text/event-stream" \
  -d '{"jsonrpc":"2.0","id":1,"method":"tools/call",
       "params":{"name":"whoami","arguments":{}}}'

A token acts as you, on every project you can reach, until you revoke it. It doesn't expire. Treat it like a password, keep it out of your repository, and revoke it under Tokens when you no longer need it.

Other agents

Any agent that speaks MCP over HTTP can connect. Point it at https://mcp.appentic.com/mcp. It discovers how to sign you in from the server, so there's nothing else to configure. Claude Code is the one we test against.

If it won't connect