Cloudflare Workers

Cloudflare Workers enable you to deploy serverless code across the globe. The Workers platform also comes with additional runtime APIs, such as Durable Objects & KV that allow you to build powerful applications. See the API documentation for more information.

Getting Started#

Before getting started, you need to install the Cloudflare wrangler CLI globally.

Next, either use the edge CLI command or setup manually:

Open the lib/main.dart file to view the minimal HTTP request fetch handler.

Each incoming request to your edge function will be passed to the fetch handler, which is expected to return a Response instance. The fetch handler provides a Request instance which contains useful information about the incoming request, such as the method, URL, headers, body and more.

Alongside the Request instance, you are also provided an Environment instance, enables you to access service bindings (e.g. Durable Objects). Additionally the ExecutionContext argument enables you to perform actions such as defining functions which can run after a response has been sent to the client.

Local development#

To run your application locally, use the wrangler dev --local command from the root of your project:

wrangler dev --local

You can remove the --local flag to run your code directly on the Cloudflare Worker environment. This is useful for testing the production environment locally, however it will take longer to start up as your code is uploaded.

This command will start a development server where you can access your application, by default at Any changes you make to your Dart code will trigger a hot-reload.


To deploy your application to Cloudflare Workers, first remove the --dev flag from the wrangler.toml file:

- command = "dart run edge build cloudflare --dev"
+ command = "dart run edge build cloudflare"

Removing the --dev flag will ensure that your application is built in production mode, which will result in a smaller bundle size and optimized code.

Please upvote this feature request to help remove this manual step!

Next, run the wrangler publish command from the root of your project:

wrangler publish