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

# Other clients

> The generic setup for any MCP client: an address and a header with a token.

Checked on Izumi 0.5.7, 2026-11-20 (the generic setup follows the server's documentation, no other client was tried live).

> [!WARNING]
>
> Not checked live: Cursor, Windsurf, Cline, VS Code, Zed, Continue and every other client except those on the ChatGPT, Claude, Slack and Claude Code pages. This page describes what Izumi expects from a client. For menus and field names in a specific client, follow that client's own documentation.

Every client that speaks MCP over Streamable HTTP is set up the same way.

## What you need

* The server address: the address of your Izumi with `/mcp` at the end. On the demo installation (invitation only): `https://izumi.akiaconsulting.com/mcp`.
* A personal token. Make it in Izumi: **My tokens and profile**, section **Tokens**, button **New token**. The steps are on the [Claude Code](https://docs.izumi.akiaconsulting.com/en/ai-tools/claude-code.md) page, section "Path 2: personal token".

The token acts as you. It never reads more than you can read, and a narrowed token reads less.

## Setup

In the client add a remote MCP server with these values:

| Setting | Value |
| - | - |
| URL | `https://izumi.akiaconsulting.com/mcp` (your address with `/mcp`) |
| Transport | Streamable HTTP |
| Header | `Authorization: Bearer <token>` |

* Every request carries the header `Authorization: Bearer <token>`.
* Streamable HTTP only. A client that supports only local (stdio) servers does not work. SSE is not used.
* Optional: the client may send the headers `X-TB-Client` (the client's name) and `X-TB-Agent` (the name of the agent running in it). Editors who review your proposals see them, and they show in your **Access log**, so an editor knows where a proposal came from.

## Check

Ask the client to call `whoami`. The answer is your name and the companies and domains the token may read. If you get an authentication error, the token is wrong, expired or revoked: make a new one.

## Limits

* The token expires at the expiry you chose, at most 90 days.
* Calls are limited to 60 per minute per token. The tool then returns `rate_limited` with the wait time.
* Revoking the token on **My tokens and profile** stops the client at once.

## Troubleshooting

* An authentication error: the token is wrong, expired or revoked. Make a new one.
* `forbidden_scope`: the token lacks the `propose` scope. Make a new token with that scope.
* `rate_limited`: wait for the time the tool names.


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