Part 2: Add a Worker

In Part 1 you deployed an R2 Bucket. Now you’ll create a Cloudflare Worker that reads and writes objects in that bucket over HTTP.

Create the Worker file

Create src/worker.ts. A Worker is a special kind of Resource — it has both an infrastructure definition and a runtime implementation expressed as an Effect.

src/worker.ts

import * as Cloudflare from "alchemy/Cloudflare";
import * as Effect from "effect/Effect";
import * as HttpServerResponse from "effect/unstable/http/HttpServerResponse";

export default Cloudflare.Worker(
  "Worker",
  {
    main: import.meta.url,
  },
  Effect.gen(function* () {
    return {
      fetch: Effect.gen(function* () {
        return HttpServerResponse.text("Hello, world!");
      }),
    };
  }),
);

Note

Notice the “Effect returning an Effect” pattern. The outer generator is the Init phase — it runs at both plan time and runtime. The inner fetch generator is the Runtime phase — it runs only when handling a request.

Bind the Bucket to the Worker

Now let’s bind the R2 Bucket from Part 1 to our new Worker. The problem is that the Bucket is declared inside the Stack’s generator in alchemy.run.ts — we can’t import it from there.

A common pattern in Alchemy is to give each resource its own file. Create src/bucket.ts:

src/bucket.ts

import * as Cloudflare from "alchemy/Cloudflare";

export const Bucket = Cloudflare.R2.Bucket("Bucket");

Update alchemy.run.ts to import it instead of declaring it inline:

alchemy.run.ts

import * as Alchemy from "alchemy";
import * as Cloudflare from "alchemy/Cloudflare";
import * as Effect from "effect/Effect";
import { Bucket } from "./src/bucket.ts";

export default Alchemy.Stack(
  "MyApp",
  {
    providers: Cloudflare.providers(),
    state: Cloudflare.state(),
  },
  Effect.gen(function* () {
    const bucket = yield* Cloudflare.R2.Bucket("Bucket");
    const bucket = yield* Bucket;

    return {
      bucketName: bucket.bucketName,
    };
  }),
);

Tip

Resource declarations like Cloudflare.R2.Bucket("Bucket") are just descriptions — they don’t execute until you yield* them inside a Stack. You can safely import them from multiple files.

Now the Worker can import Bucket and bind it in the Init phase:

import * as Cloudflare from "alchemy/Cloudflare";
import * as Effect from "effect/Effect";
import * as HttpServerResponse from "effect/unstable/http/HttpServerResponse";
import { Bucket } from "./bucket.ts";

export default Cloudflare.Worker(
  "Worker",
  {
    main: import.meta.url,
  },
  Effect.gen(function* () {
Error ts(2345)  ― Type 'ReadWriteBucket' is not assignable to type 'WorkerServices | PlatformServices'.
    const bucket = yield* Cloudflare.R2.ReadWriteBucket(Bucket);

    return {
      fetch: Effect.gen(function* () {
        return HttpServerResponse.text("Hello, world!");
      }),
    };
  }),
);

Provide the binding layer

The previous step showed a type error — Cloudflare.R2.ReadWriteBucket(Bucket) requires the Cloudflare.R2.ReadWriteBucketBinding service. Fix it by piping the outer Effect through Cloudflare.R2.ReadWriteBucketBinding:

import * as Cloudflare from "alchemy/Cloudflare";
import * as Effect from "effect/Effect";
import * as HttpServerResponse from "effect/unstable/http/HttpServerResponse";
import { Bucket } from "./bucket.ts";

export default Cloudflare.Worker(
  "Worker",
  { main: import.meta.url },
  Effect.gen(function* () {
    const bucket = yield* Cloudflare.R2.ReadWriteBucket(Bucket);

    return {
      fetch: Effect.gen(function* () {
        return HttpServerResponse.text("Hello, world!");
      }),
    };
  }).pipe(Effect.provide(Cloudflare.R2.ReadWriteBucketBinding)),
);

ReadWriteBucketBinding tells the Worker runtime how to look up the underlying R2 binding from the Cloudflare environment. Without it, the ReadWriteBucket binding wouldn’t know where to find the bucket at runtime.

Add the PUT handler

Let’s replace the placeholder response with a PUT route that stores objects in the bucket. Add HttpServerRequest to access the incoming request:

import * as Cloudflare from "alchemy/Cloudflare";
import * as Effect from "effect/Effect";
import { HttpServerRequest } from "effect/unstable/http/HttpServerRequest";
import * as HttpServerResponse from "effect/unstable/http/HttpServerResponse";
import { Bucket } from "./bucket.ts";

export default Cloudflare.Worker(
  "Worker",
  { main: import.meta.url },
  Effect.gen(function* () {
    const bucket = yield* Cloudflare.R2.ReadWriteBucket(Bucket);
    return {
      fetch: Effect.gen(function* () {
        const request = yield* HttpServerRequest;
        const key = request.url.split("/").pop()!;

        if (request.method === "PUT") {
          yield* bucket.put(key, request.stream, {
            contentLength: Number(request.headers["content-length"] ?? 0),
          });
          return HttpServerResponse.empty({ status: 201 });
        }

        return HttpServerResponse.text("Hello, world!");
      }),
    };
  }).pipe(Effect.provide(Cloudflare.R2.ReadWriteBucketBinding)),
);

TypeScript flags a type error. The bucket.put call can fail with R2Error, but a Worker’s fetch handler only allows HttpServerError or HttpBodyError. Effect tracks this in the type system — you can’t forget to handle it.

Handle R2 errors

Pipe the fetch Effect through Effect.catchTag to convert R2Error into a 500 response:

import * as Cloudflare from "alchemy/Cloudflare";
import * as Effect from "effect/Effect";
import { HttpServerRequest } from "effect/unstable/http/HttpServerRequest";
import * as HttpServerResponse from "effect/unstable/http/HttpServerResponse";
import { Bucket } from "./bucket.ts";

export default Cloudflare.Worker(
  "Worker",
  { main: import.meta.url },
  Effect.gen(function* () {
    const bucket = yield* Cloudflare.R2.ReadWriteBucket(Bucket);
    return {
      fetch: Effect.gen(function* () {
        const request = yield* HttpServerRequest;
        const key = request.url.split("/").pop()!;

        if (request.method === "PUT") {
          yield* bucket.put(key, request.stream, {
            contentLength: Number(request.headers["content-length"] ?? 0),
          });
          return HttpServerResponse.empty({ status: 201 });
        }

        return HttpServerResponse.text("Hello, world!");
      }).pipe(
        Effect.catchTag("R2Error", (error) =>
          Effect.succeed(
            HttpServerResponse.text(error.message, { status: 500 }),
          ),
        ),
      ),
    };
  }).pipe(Effect.provide(Cloudflare.R2.ReadWriteBucketBinding)),
);

Effect.catchTag matches errors by their _tag field. If an R2 operation fails at runtime, the Worker returns a 500 instead of crashing — and the type error disappears because R2Error is now fully handled.

Add the GET handler

Complete the fetch handler by reading objects from the bucket when the request isn’t a PUT:

import * as Cloudflare from "alchemy/Cloudflare";
import * as Effect from "effect/Effect";
import { HttpServerRequest } from "effect/unstable/http/HttpServerRequest";
import * as HttpServerResponse from "effect/unstable/http/HttpServerResponse";
import { Bucket } from "./bucket.ts";

export default Cloudflare.Worker(
  "Worker",
  { main: import.meta.url },
  Effect.gen(function* () {
    const bucket = yield* Cloudflare.R2.ReadWriteBucket(Bucket);
    return {
      fetch: Effect.gen(function* () {
        const request = yield* HttpServerRequest;
        const key = request.url.split("/").pop()!;

        if (request.method === "PUT") {
          yield* bucket.put(key, request.stream, {
            contentLength: Number(request.headers["content-length"] ?? 0),
          });
          return HttpServerResponse.empty({ status: 201 });
        }

        const object = yield* bucket.get(key);
        if (object === null) {
          return HttpServerResponse.text("Not found", { status: 404 });
        }
        const text = yield* object.text();
        return HttpServerResponse.text(text);
        return HttpServerResponse.text("Hello, world!");
      }).pipe(
        Effect.catchTag("R2Error", (error) =>
          Effect.succeed(
            HttpServerResponse.text(error.message, { status: 500 }),
          ),
        ),
      ),
    };
  }).pipe(Effect.provide(Cloudflare.R2.ReadWriteBucketBinding)),
);

The Worker now handles two routes:

• PUT /:key — stores the request body in the bucket
• GET /:key — retrieves the object, returning 404 if missing

Because bucket.get also returns R2Error, the catchTag you added in the previous step already covers it — no additional error handling needed.

Wire the Worker into the Stack

Add the Worker to alchemy.run.ts and expose its URL as a stack output:

import * as Alchemy from "alchemy";
import * as Cloudflare from "alchemy/Cloudflare";
import * as Effect from "effect/Effect";
import { Bucket } from "./src/bucket.ts";
import Worker from "./src/worker.ts";

export default Alchemy.Stack(
  "MyApp",
  {
    providers: Cloudflare.providers(),
    state: Cloudflare.state(),
  },
  Effect.gen(function* () {
    const bucket = yield* Bucket;
    const worker = yield* Worker;

    return {
      bucketName: bucket.bucketName,
      url: worker.url,
    };
  }),
);

Deploy

Deploy again. Alchemy detects the new Worker and the unchanged Bucket:

bun

bun alchemy deploy

npm

npm run alchemy deploy

pnpm

pnpm alchemy deploy

yarn

yarn alchemy deploy

Plan: 1 to create

+ Worker (Cloudflare.Worker) (1 bindings)
  + Bucket
• Bucket (Cloudflare.R2.Bucket)

Proceed?
◉ Yes ○ No
• Bucket (Cloudflare.R2.Bucket) no change
✓ Worker (Cloudflare.Worker) created
  • Uploading worker (14.20 KB) ...
  • Enabling workers.dev subdomain...
{
  bucketName: "myapp-bucket-a1b2c3d4e5",
  url: "https://myapp-worker-dev-you-abc123.workers.dev",
}

Try it out

Use curl to write and read an object:

# Store an object
curl -X PUT https://myapp-worker-dev-you-abc123.workers.dev/hello.txt \
  -d 'Hello, world!'

# Retrieve it
curl https://myapp-worker-dev-you-abc123.workers.dev/hello.txt
# → Hello, world!

Recap

You now have:

• A Cloudflare Worker with GET and PUT routes
• An R2 Bucket bound to the Worker
• Stack outputs showing both the bucket name and worker URL

In Part 3, you’ll learn about stages and state stores so multiple developers (and CI) can deploy isolated environments.

Want to start building now?

You have a deployed Worker — that’s enough to skip ahead and add a Durable Object (or any other Cloudflare deep-dive) if you’d rather build something interesting first and circle back to testing, local dev, and CI/CD when you need them.