Skip to main content

Documentation Index

Fetch the complete documentation index at: https://mintlify.com/eersnington/sideffect/llms.txt

Use this file to discover all available pages before exploring further.

WorkflowEntrypoints is the Cloudflare-specific adapter that converts Sideffect WorkflowLayer values into native Cloudflare WorkflowEntrypoint subclass constructors. Use it in a plain Cloudflare Worker — without Vite — when you want to wire up Sideffect workflows manually. Import it from sideffect/cloudflare.
If you are using Vite, you do not need to call WorkflowEntrypoints.make() yourself. withCloudflareWorkflows() generates a virtual entry module that calls it automatically for every discovered layer. See the withCloudflareWorkflows API reference for the automatic Vite integration.

​
WorkflowEntrypoints.make()

​
Function signature

WorkflowEntrypoints.make<const Entries extends WorkflowLayerEntries>(
  entries: Entries
): { readonly [K in keyof Entries]: WorkflowEntrypointConstructor }

​
Parameters

​
entries
WorkflowLayerEntries
required
A plain object whose keys are class_name strings — the Cloudflare Workflow class names that appear in your Wrangler config — and whose values are WorkflowLayer instances created by Workflow.make(...).toLayer(...).Each key becomes a named export that Cloudflare’s runtime uses to route incoming workflow invocations. Keys must be valid JavaScript identifiers (the same rule Cloudflare enforces for WorkflowEntrypoint class names in Wrangler config).
{
  ResizeImage: resizeImageLayer,
  SendWelcomeEmail: sendWelcomeEmailLayer,
}

​
Return value

​
entrypoints
object
An object with the same keys as entries. Each value is a WorkflowEntrypoint subclass constructor. Destructure and export the keys whose names match the class_name values in your Wrangler workflow bindings.
export const { ResizeImage } = WorkflowEntrypoints.make({ ResizeImage: resizeImageLayer });

​
Validation

WorkflowEntrypoints.make() validates every entry before returning and throws synchronously on the first violation:
  1. Invalid class_name identifier — if a key is not a valid JavaScript identifier (tested against /^[$A-Z_a-z][$\w]*$/), a TypeError is thrown: 
    TypeError: Invalid Cloudflare Workflow class_name "resize-image". Sideffect generates a named
JavaScript export for each workflow, so class_name must be a valid identifier such as
"ResizeImage". Update the Wrangler workflow class_name and the matching Worker export.
  2. Value is not a WorkflowLayer — if a value does not have _tag === "WorkflowLayer", a TypeError is thrown: 
    TypeError: Expected Worker export "ResizeImage" to be a Sideffect WorkflowLayer. Export a layer
with the same name as the Wrangler workflow class_name, for example:
export { resizeImageWorkflowLayer as ResizeImage } from "./workflows/resize-image".

​
Examples

​
Single workflow

import { WorkflowEntrypoints } from "sideffect/cloudflare";
import { resizeImageLayer } from "./workflows/resize-image";

export const { ResizeImage } = WorkflowEntrypoints.make({
  ResizeImage: resizeImageLayer,
});
Your wrangler.toml should have a matching binding: 
[[workflows]]
binding = "RESIZE_IMAGE"
name    = "resize-image"
class_name = "ResizeImage"

​
Multiple workflows

import { WorkflowEntrypoints } from "sideffect/cloudflare";
import { resizeImageLayer } from "./workflows/resize-image";
import { sendWelcomeEmailLayer } from "./workflows/send-welcome-email";

export const { ResizeImage, SendWelcomeEmail } = WorkflowEntrypoints.make({
  ResizeImage: resizeImageLayer,
  SendWelcomeEmail: sendWelcomeEmailLayer,
});

​
Re-exporting with a different local name

When your layer variable name differs from the required class_name, rename it at the export site rather than renaming the layer itself: 
import { WorkflowEntrypoints } from "sideffect/cloudflare";
import { imageResizeWorkflowLayer } from "./workflows/resize-image";

export const { ResizeImage } = WorkflowEntrypoints.make({
  ResizeImage: imageResizeWorkflowLayer,
});

​
Runtime behaviour

At workflow invocation time Cloudflare calls the run(event, step) method on the generated WorkflowEntrypoint subclass. Internally the generated class delegates to WorkflowEngine.run(), which:
  1. Decodes the payload — runs the workflow’s payloadSchema against event.payload. If decoding fails the error is converted to a NonRetryableError so Cloudflare does not retry the workflow with a payload it cannot parse.
  2. Creates a SideffectStep façade — wraps Cloudflare’s native step object with typed step.do(), step.sleep(), step.sleepUntil(), and step.waitForEvent() methods that handle payload/result schema decoding for each step.
  3. Calls the layer’s run function — passes the decoded payload, the WorkflowContext (containing env and ctx), and the SideffectStep façade.
  4. Converts NonRetryableError — if the layer’s run function throws a Sideffect NonRetryableError, it is re-thrown as the native Cloudflare NonRetryableError from cloudflare:workflows so the platform stops retrying immediately.

Build docs developers (and LLMs) love

Get started for freeTalk to us