Logto Client SDK
The Logto JavaScript Client SDK written in TypeScript.
Installation
Using npm
npm install @logto/client
Using yarn
yarn add @logto/client
Using pnpm
pnpm add @logto/client
What is this and how does it work?
Logto JavaScript Client SDK is platformless, and is the foundation of the other platform's SDKs (Browser, Next.js, React, Vue, etc.). Usually you are not expected to use it directly in your application, as we have released a set of official SDKs to help you integrate Logto with your favorite JavaScript frameworks. Check this out and get started!
If Logto does not support your framework and you want to contribute by building a new SDK, we recommend checking out our Browser SDK and Node.js SDK and start from there.
Adapters
To implement a platform-specific SDK, you should implement the following adapters:
- storage: save tokens and other info.
- navigate: handle redirect.
- generateState: generate state.
- generateCodeVerifier: generate code verifier.
- generateCodeChallenge: generate code challenge.
- cache: optionally cache well-known data.
The client uses the runtime's native fetch by default. Set the optional fetch adapter only when
the platform needs a custom transport.
The optional cache adapter stores well-known data. Cache reads and writes are best-effort and do
not make an otherwise successful request fail. The deprecated unstable_cache property remains
supported; cache takes precedence when both are provided.
See the adapters for more information.
Request timeouts
Set requestTimeoutMs in LogtoConfig to apply a timeout to each request to the Logto server. The
option covers discovery, authorization-code and refresh-token exchange, revocation, user info, and
remote JSON Web Key Set requests. Framework SDKs built on @logto/client inherit the same option.
const config = {
endpoint: 'https://example.logto.app',
appId: 'your-app-id',
requestTimeoutMs: 10_000,
};
The timeout is opt-in. The client builds its requester on top of native fetch, or the adapter's
custom fetch transport when provided, so the same timeout policy applies in both cases. A custom
transport must observe RequestInit.signal and settle after the signal is aborted. The client
awaits that settlement before it rejects the operation.
The deprecated requester adapter remains supported for compatibility. When both fetch and
requester are provided, fetch takes precedence. Since a custom requester is already fully
configured, the client does not apply requestTimeoutMs to it.