---
title: API tokens and client keys
---

For Unleash to be of any use, it requires at least a server and a [consuming client](../reference/sdks/index.md). More advanced use cases may call for multiple clients, automated feature toggle updates, the [Unleash proxy](../generated/unleash-proxy.md) and [Unleash proxy clients](../reference/sdks/index.md#front-end-sdks), and more. To facilitate communication between all these moving parts, Unleash uses a system of API tokens and client keys, all with a specific purpose in mind.

This document details the three kinds of tokens and keys that you will need to fully connect any Unleash system:

-   [Admin tokens](#admin-tokens) for updating resources in Unleash
-   [Client tokens](#client-tokens) for connecting server-side client SDKs and the Unleash proxy to the Unleash server
-   [Proxy client keys](#proxy-client-keys) for connecting proxy client SDKs to the Unleash proxy.

## API tokens

:::tip

This section describes what API tokens are. For information on how to create them, refer to the [how-to guide for creating API tokens](../how-to/how-to-create-api-tokens.mdx).

:::

Use API tokens to connect to the Unleash server API. API tokens come in four distinct types:

-   [Admin tokens](#admin-tokens)
-   [Personal access tokens](#personal-access-tokens)
-   [Client tokens](#client-tokens)
-   [Front-end tokens](#front-end-tokens)

All types use [the same format](#format) but have different intended uses. Admin and client tokens are _secrets_ and should _not_ be exposed to end users. Front-end tokens, on the other hand, are not secret.

### The parts of an API token {#token-data}

Admin, client and front-end tokens contain the following pieces of information:

| Name | Description |
| --- | --- |
| Token name (sometimes called "username") | The token's name. Names are **not** required to be unique. |
| Type | What kind of token it is: admin, client, or front-end. |
| Projects | What projects a token has access to. |
| Environment | What environment the token has access to. |

Personal access tokens follow their own special format, and only contain an optional description for the token and an expiry date.

### Admin tokens

**Admin tokens** grant _full read and write access_ to all resources in the Unleash server API. Admin tokens have access to all projects, all environments, and all global resources (find out more about [resources in the RBAC document](../reference/rbac.md#core-principles)).

Use admin tokens to:

-   Automate Unleash behavior such as creating feature toggles, projects, etc.
-   Write custom Unleash UIs to replace the default Unleash admin UI.

Do **not** use admin tokens for:

-   [Client SDKs](../reference/sdks/index.md): You will _not_ be able to read toggle data from multiple environments. Use [client tokens](#client-tokens) instead.

Support for scoped admin tokens with more fine-grained permissions is currently in the planning stage.

### Personal access tokens

**Personal access tokens** are a special form of admin tokens and grant access to the same resources that the user that created them has access to. These permissions are dynamic, so if a user's permissions change through addition of a custom role, the token will likewise have altered permissions.

When using a personal access token to modify resources, the event log will list the token creator's name for that operation.

Personal access tokens with a lifetime **will stop working after the expiration date**.

Use personal access tokens to:

- Provide more fine-grained permissions for automation than an admin token provides
- Give temporary access to an automation tool

:::info On token expiration

It is possible to set a token's expiration date to **never**. However, a token that doesn't expire brings with it a few security concerns. We recommend that you use tokens with expiration dates whenever possible.

:::

Do **not** use personal access tokens for:

-   [Client SDKs](../reference/sdks/index.md): You will _not_ be able to read toggle data from multiple environments. Use [client tokens](#client-tokens) instead.
-   Write custom Unleash UIs: Personal access tokens may expire and their permissions may change. It's better to use [admin tokens](#admin-tokens) tokens instead.

### Client tokens

**Client tokens** are intended for use in [server-side client SDKs](../reference/sdks/index.md#server-side-sdks) (including the Unleash Proxy) and grant the user permissions to:

-   Read feature toggle information
-   Register applications with the Unleash server
-   Send usage metrics

When creating a client token, you can choose which projects it should be able to read data from. You can give it access to a specific list of projects or to all projects (including projects that don't exist yet). Prior to Unleash 4.10, a token could be valid only for a _single project_ or _all projects_.

Each client token is only **valid for a single environment**.

Use client tokens:

-   In [server-side client SDKs](../reference/sdks/index.md#server-side-sdks)
-   To connect [the Unleash Proxy](../generated/unleash-proxy.md) to the Unleash API

Do **not** use client tokens in:

-   [Front-end SDKs](../reference/sdks/index.md#front-end-sdks). You will _not_ be able to connect to the Unleash server due to CORS restrictions. To connect front-end SDKs, choose one of the following options:
-   Enable the [Unleash front-end API](./front-end-api.md) and create a [front-end token](#front-end-tokens).
-   Configure an [Unleash Proxy](../generated/unleash-proxy.md) and use [Proxy client keys](#proxy-client-keys).

### Front-end tokens

**Front-end tokens** are used with [front-end SDKs](../reference/sdks/index.md#front-end-sdks) when used with the [Unleash front-end API](./front-end-api.md). They grant the user permission to:

-   Read the enabled toggled for a given context
-   Register applications with the Unleash server
-   Send usage metrics

As with [client tokens](#client-tokens), front-end tokens can read data from one, multiple, or all existing projects.

Each front-end token is only **valid for a single environment**.

Use front-end tokens in:

-   [Front-end SDKs (also known as _proxy clients_)](../reference/sdks/index.md#front-end-sdks).

Do **not** use front-end tokens in:

-   [Server-side SDKs](../reference/sdks/index.md#server-side-sdks). The format is different, so they won't work correctly.

### Format

API tokens come in one of two formats. When we introduced [environments](./environments.md) in Unleash 4.3, we updated the format of the tokens to provide more human-readable information to the user. Both formats are still valid (you don't need to update a working token of the old format) and are described below.

#### Version 1

The first version of API tokens was a 64 character long hexadecimal string. Example:

```
be44368985f7fb3237c584ef86f3d6bdada42ddbd63a019d26955178
```

#### Version 2

API tokens consist of three parts:

1. Project(s)
2. Environment
3. Hash

The parts are separated by two different separators: A colon (`:`) between the project(s) and the environment, and a full stop (`.`) between the environment and the hash.

The **project(s)** part is one of:

-   The id of a specific project, for example: `default`. This indicates that the token is **only valid for this project**.
-   A pair of opening and closing square brackets: `[]`. This indicates that the token is **valid for a discrete list of projects**. The list of projects is not shown in the token.
-   An asterisk: `*`. This indicates that the token is **valid for all projects (current and future)**.

The **environment** is the name of an environment on your Unleash server, such as `development`.

The **hash** is 64 character long hexadecimal string.

Personal access tokens do not contain project or environment information, since they mimic the user that created them. Instead, the token starts with the string `user`.

Some example client tokens are:

-   A token with access to toggles in the "development" environment of a single project, "project-a":
    ```
    project-a:development.be44368985f7fb3237c584ef86f3d6bdada42ddbd63a019d26955178
    ```
-   A token with access to toggles in the "production" environment multiple projects:
    ```
    []:production.be44368985f7fb3237c584ef86f3d6bdada42ddbd63a019d26955178
    ```
-   A token with access to toggles in the "development" environment of all projects:
    ```
    *:development.be44368985f7fb3237c584ef86f3d6bdada42ddbd63a019d26955178
    ```
-   A personal access token:
    ```
    user:be7536c3a160ff15e3a92da45de531dd54bc1ae15d8455c0476f086b
    ```

## Proxy client keys {#proxy-client-keys}

Use proxy client keys to connect [Proxy client SDKs (front-end SDKs)](../reference/sdks/index.md#front-end-sdks) to the [Unleash Proxy](../generated/unleash-proxy.md). As opposed to the [API tokens](#api-tokens), Proxy client keys are _not_ considered secret and are safe to use on any clients (refer to the [the proxy documentation for more about privacy](../generated/unleash-proxy.md#we-care-about-privacy)). They do _not_ let you connect to the Unleash server API.

Proxy client keys are arbitrary strings that you _must_ provide the Unleash proxy with on startup. They can be whatever you want and you **create them yourself**.

:::info Creating proxy client keys

To designate a string as a proxy client key, add it to the `clientKeys` list when starting the proxy, as mentioned in the [_configuration_ section of the Unleash proxy documentation](../generated/unleash-proxy.md#configuration). Connecting clients should then specify the same string as their client key.

:::

Unleash does not generate proxy client keys for you. Because of this, they have no specific format.

Use Proxy client keys to:

-   Connect [Proxy client SDKs](../reference/sdks/index.md#front-end-sdks) to the [Unleash Proxy](../generated/unleash-proxy.md)
-   Connect your own custom Proxy clients (or pure HTTP requests) to the Unleash Proxy

Do **not** use Proxy client keys to:

-   Connect to the Unleash API. It will not work. Use an appropriate [API token](#api-tokens) instead.