---
title: "MCP Server for Your Apps and APIs - APIANT.ai"
description: "Connect your AI client to your apps through one remote MCP server. Sign in once in your browser; your agent builds, tests and runs integrations."
canonical: "https://apiant.ai/mcp"
---

# One MCP connection to your apps.

APIANT runs a remote MCP server at `https://mcp.apiant.ai`. Add that address to your AI client as a remote MCP server and sign in once in your browser. There is nothing to install. Your agent can then search the app catalog, connect your accounts, build connectors from an app's API documentation, and build, test and run automations, all by asking.

[Start free](https://app.apiant.ai/register?utm_source=mcp&utm_medium=cta&utm_campaign=start_free&utm_content=hero)[Read the MCP docs](https://apiant.ai/docs/mcp)

## Set it up

### Copy the address

Open **Connect your AI client** from the account menu in APIANT.ai. The address and a ready-made Claude Code terminal command are there.

### Add it to your AI client

Paste the address as a remote MCP server, or run the Claude Code command. [Step by step for Claude Code](https://apiant.ai/docs/mcp/claude-code).

### Sign in once

Your browser opens APIANT's sign-in, including multi-factor authentication if you use it. The client gets its own OAuth token; your password never reaches it. [Authentication and permissions](https://apiant.ai/docs/mcp/authentication).

## What your agent can do over MCP

### Find and connect apps

Search the [catalog](https://apiant.ai/apps), see each app's triggers and actions, and connect your accounts.

### Build a missing connector

Given an app's name, the agent reads the vendor's API documentation, builds the connector and its authentication, and tests each operation against the live API. [How connector building works](https://apiant.ai/docs/apps/custom-connectors).

### Build and test automations

Describe an automation; the agent builds it, tests every scenario, fixes failures and turns it on.

### Run and diagnose

Read execution history, find what failed today, retry runs and restart them from a step.

### Publish and deploy

[Publish](https://apiant.ai/docs/account/publish-and-deploy) from a dev environment to production, or deploy an automation into other accounts.

### Expose automations as tools

Turn an automation into a tool your AI client can call, so a workflow becomes one step in a larger agent task.

## Any API, not a fixed list

The catalog is where your agent starts, not where it stops. It can extend the catalog for your account: point it at the API documentation and it builds the operations it needs, at whatever depth the API allows. If the documentation cannot be read, it stops and says so rather than guessing endpoints.

## Or add the Claude plugin

In Claude Code, the APIANT plugin sets up the same connection: add the marketplace with `/plugin marketplace add APIANT/apiant-plugins`, then `/plugin install apiant@apiant`, and sign in with your APIANT account. The plugin and its manifest are public on [GitHub](https://github.com/APIANT/apiant-plugins).

## Listed in the MCP Registry

The server is published in the official MCP Registry as `ai.apiant/apiant`, verified against the apiant.ai domain: [see the registry entry](https://registry.modelcontextprotocol.io/v0/servers?search=ai.apiant/apiant). It is also listed on [Glama](https://glama.ai/mcp/connectors/ai.apiant/apiant). Its server card is at [/.well-known/mcp/server-card.json](https://apiant.ai/.well-known/mcp/server-card.json).

## Skills travel with the connection

The procedures your agent follows, for building connectors, testing automations or diagnosing a failed run, are delivered over the MCP connection itself, so every client that connects gets the current version. [See the skills](https://apiant.ai/docs/reference/skills) and [the MCP tools](https://apiant.ai/docs/reference/mcp-tools). The server's OAuth scopes and the public app catalog feed are also described in an [OpenAPI spec](https://apiant.ai/openapi.json).

## Questions

### What is the APIANT MCP server?

A remote MCP server at https://mcp.apiant.ai that exposes your APIANT.ai account as tools for your AI client: the app catalog, connections, connector building, automations, runs and alerts.

### Do I need to install anything?

No. You add the server address to your AI client and sign in once in your browser.

### How does authentication work?

OAuth with PKCE. Your browser opens APIANT's sign-in, including multi-factor authentication, and the client receives its own token. You can disconnect it at any time.

### Which AI clients can I use?

Setup is documented step by step for Claude Code. The server uses the standard remote MCP transport (Streamable HTTP) with OAuth sign-in.

### Does using it over MCP cost credits?

No. Assistant credits apply to the in-app assistant. Automation runs, including test runs that act on your apps, count toward your plan's monthly tasks.

### Can it connect an app that is not in the catalog?

Yes. The agent builds the connector from the vendor's API documentation and tests each operation against the live API before you use it.

## Describe the first one.

Free for everyone. There is no dashboard to learn: your agent is the interface.

[Start free](https://app.apiant.ai/register?utm_source=mcp&utm_medium=cta&utm_campaign=start_free&utm_content=closing)[Browse the apps](https://apiant.ai/apps)