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

# Terminal authentication is stabilized

> Clients can run an Agent's interactive login flow in a terminal in ACP v1 and the v2 baseline.

**Published:** August 20, 2026

The [Terminal Authentication RFD](https://agentclientprotocol.com/rfds/auth-methods.md) has moved to Completed. An Agent can advertise an authentication method with `type: "terminal"` for a login flow that needs its terminal UI. The Client launches the Agent program it already has configured in an interactive terminal, with the `args` and `env` the method supplies, and the user logs in there. The method cannot supply a command. Exit status zero means the login succeeded. The terminal process is separate from the ACP connection, so the Client does not call `authenticate` or `auth/login` for a `terminal` method.

In [ACP v1](https://agentclientprotocol.com/protocol/v1/authentication.md#terminal-authentication), Clients opt in with `clientCapabilities.auth.terminal: true` and reconnect to the Agent after a successful login. In [ACP v2](https://agentclientprotocol.com/protocol/v2/authentication.md#terminal-authentication), Clients opt in with `capabilities.auth.terminal: {}`. After a successful login they reconnect and resume their sessions with `session/resume`, or first retry on the existing connection.


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