Task Library

These are the built-in tasks that are available to use in your Jobs.

Welcome to the Trigger.dev Task Library 📚. You may be wondering, what is a Task and why are there a library of them? Well you see, Trigger.dev works by divvying up a long-running job execution into a bunch of little tasks, each one taking less time then a single serverless function execution. 💫

You can define and run your own tasks easily using io.runTask(), or you can use one of our Integrations which are tasks for specific APIs, like OpenAI or Stripe.

Read more about how Tasks work here.

We also have a growing library of built-in tasks that you can use in your Jobs through the io object. These tasks are designed to be generic and reusable, and are a great way to get started with Trigger.dev.

You may notice that I’m using emojis for all the cache keys below, which is totally 💯% fine as long as they are unique inside a run. Read more about how cache keys work here

wait

This task allows you to resume executing your job after a certain amount of time has passed:

await io.wait("⏰", 60); // wait 60 seconds

Internally this task is considered a “noop”, and noop tasks have no output.

reference docs

waitForRequest

You supply this task with a callback to receive a URL. When a POST request is made to that URL, the JSON body of the request becomes the task output.

The example below uses waitForRequest to capture a Screenshot of a website using ScreenshotOne.com and passes the callback URL to the webhook URL to get notified when the screenshot is finished:

const result = await io.waitForRequest<ScreenshotResponse>(
  "📸",
  async (url) => {
    await fetch(`https://api.screenshotone.com/take`, {
      method: "POST",
      headers: {
        "Content-Type": "application/json",
      },
      body: JSON.stringify({
        webhook_url: url, // this is the URL that will be called when the screenshot is ready
        access_key: "my-access-key",
        url: "https://trigger.dev",
        store: "true",
        storage_path: "my-screeshots",
        response_type: "json",
        async: "true",
        storage_return_location: "true",
      }),
    });
  },
  {
    timeoutInSeconds: 300, // wait up to 5 minutes for the screenshot to be ready
  }
);

We actually originally built this task for our Replicate integration, which accepts a callback URL to notify you when a prediction is ready. So this allows you to write very succinct code to create a prediction and wait for it’s results:

const sdPrediction = await io.replicate.predictions.createAndAwait("🧑‍🎨", {
  version: "ac732df83cea7fff18b8472768c88ad041fa750ff7682a21affe81863cbe77e4",
  input: {
    prompt: "What is the meaning of life?",
  },
});

reference docs

waitForEvent

This task allows you to wait for an event to be sent. To read about how events work, check out the Events documentation.

const event = await io.waitForEvent(
  "🥂",
  {
    name: "user.created",
    schema: z.object({
      id: z.string(),
      createdAt: z.coerce.date(),
      isAdmin: z.boolean(),
    }),
    filter: {
      isAdmin: [true], // Only wait for events where isAdmin is true
    },
  },
  {
    timeoutInSeconds: 60 * 60, // Wait for up to an hour
  }
);

The event object returned from this task is the full event object that was sent, including id, name, payload, context, and more.

reference docs

backgroundFetch

This task allows you to perform a fetch request in the background, and then resume the execution of your job after the request has completed.

const body = io.backgroundFetch<MyResponseData>("🕸️", "https://example.com/api/endpoint", {
  method: "POST",
  headers: {
    "Content-Type": "application/json",
    Authorization: redactString`Bearer ${auth.apiKey}`,
  },
  body: JSON.stringify({ foo: "bar" }),
});

This is useful for when an API is slow to respond and might not finish before your serverless function times out. We created this task to power our OpenAI integration, which can sometimes take more than a minute to respond:

// This uses backgroundFetch under the hood
await io.openai.chat.completions.backgroundCreate("💬", {
  model: "gpt-3.5-turbo",
  messages: [
    {
      role: "user",
      content: "Create a good programming joke about background jobs",
    },
  ],
});

reference docs

backgroundPoll

This task is similar to backgroundFetch, but instead of waiting for a single request to complete, it will poll a URL until it returns a certain value.

const result = await io.backgroundPoll<{ foo: string }>("🔃", {
  url: "https://example.com/api/endpoint",
  interval: 10, // every 10 seconds
  timeout: 300, // stop polling after 5 minutes
  responseFilter: {
    // stop polling once this filter matches
    status: [200],
    body: {
      status: ["SUCCESS"],
    },
  },
});

logger

The logger object allows you to log messages to the Trigger.dev console. This is useful for debugging your jobs, or just to see what’s going on inside your job.

await io.logger.info("This is an info message");

You can optionally pass a context object to the logger, which will be displayed in the console:

await io.logger.info("This is an info message", {
  foo: "bar",
});

We support the following log levels:

• io.logger.debug()
• io.logger.info()
• io.logger.warn()
• io.logger.error()

You may notice these tasks don’t include cache keys. We automatically create a cache key for you based on the message and the log-level

reference docs

store

The store object exposes several namespaced Key-Value Stores you can access inside of your Jobs. This is useful for storing small amounts of serializable data for later retrieval:

// store some data
await io.store.job.set("💾", "disk-A", "Doom 1.2 Demo #3");

// get it again later
const value = await io.store.job.get<string>("read-💾", "disk-A");

If you want to access the store from outside a run (e.g. just from your backend) you should use client.store instead.

The following namespaces are at your disposal:

• store.env to access and store data within the Environment
• store.job to access and store data within the Job
• store.run to access and store data within the Run

reference docs

random

Use this task to generate a random number that stays stable during run retries/resumes:

const randomNumber = await io.random("🎲", {
  min: 1,
  max: 100,
});

reference docs

sendEvent

This task allows you to send an event from inside your job run.

If you want to send an event from outside a run (e.g. just from your backend) you should use client.sendEvent() instead.

await io.sendEvent("🚚", {
  id: "e_1234567890",
  name: "new.user",
  payload: {
    userId: "u_1234567890",
  },
});

reference docs

sendEvents

This task allows you to send multiple events from inside your job run.

If you want to send multiple events from outside a run (e.g. just from your backend) you should use client.sendEvents() instead.

await io.sendEvents("🚚🚚", [
  {
    id: "e_12345",
    name: "new.user",
    payload: {
      userId: "u_12345",
    },
  },
  {
    id: "e_67890",
    name: "new.user",
    payload: {
      userId: "u_67890",
    },
  },
]);

reference docs

getEvent

This task allows you to get an event by ID from inside your job run.

If you want to get an event from outside a run (e.g. just from your backend) you should use client.getEvent() instead.

const event = await io.getEvent("📥", "e_1234567890");

reference docs

cancelEvent

If you send an event that has a delivery date in the future, you can use this task to cancel it.

await io.sendEvent(
  "🚚",
  {
    id: "e_1234567890",
    name: "new.user",
    payload: {
      userId: "u_1234567890",
    },
  },
  {
    deliverAt: new Date(Date.now() + 1000 * 60 * 60 * 24), // deliver in 24 hours
  }
);

// Later on, if you want to cancel the event:
await io.cancelEvent("🚫", "e_1234567890");

createStatus

Coming soon