> ## Documentation Index
> Fetch the complete documentation index at: https://docs.aident.ai/llms.txt
> Use this file to discover all available pages before exploring further.

# MCP setup for chat clients

> Connect a user-managed MCP client to the public Loadout endpoint when the CLI is unavailable.

Follow the [agent setup guide](/loadout/setup-agents) for a supported client. For a chat app or MCP client that uses a remote server URL, add this server in the client's app or connector settings:

```text theme={null}
https://loadout.aident.ai/mcp
```

<Steps>
  <Step title="Add the server">
    In ChatGPT, create a custom app. In Claude chat or Desktop, use Connectors if your plan supports a remote MCP URL. In another client, use its documented remote HTTP MCP configuration.
  </Step>

  <Step title="Authorize Aident">
    Complete the browser sign-in opened by the client. Grant only the scopes required by your intended work.
  </Step>

  <Step title="Enable and test">
    Enable the connection in the current conversation. Ask the agent to search Loadout for a read-only Action and check the needed Integration's status.
  </Step>
</Steps>

MCP exposes the same public Loadout concepts: Action search and metadata, Action execution, public Skill guidance, Vault connection management, and Audit. Tool availability depends on your authorization and current catalog. Use the live tool schema rather than old tool-name lists in copied configuration examples.

If authorization is interrupted, return to your client's connector settings and reconnect. For a shell-capable agent, `aident doctor` and `aident whoami` are the corresponding CLI checks. See [troubleshooting](/loadout/troubleshooting).


This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.