Code on GitHub Slack Platform Home

Web API

The @slack/web-api package contains a simple, convenient, and configurable HTTP client for making requests to Slack’s Web API. Use it in your app to call any of the over 130 methods, and let it handle formatting, queuing, retrying, pagination, and more.

Installation

$ npm install @slack/web-api

Initialize the client

The package exports a WebClient class. All you need to do is instantiate it, and you’re ready to go. You’ll typically initialize it with a token, so that you don’t have to provide the token each time you call a method. A token usually begins with xoxb or xoxp. You get them from each workspace an app is installed onto. The app configuration pages help you get your first token for your development workspace.

const { WebClient } = require('@slack/web-api');

// Read a token from the environment variables
const token = process.env.SLACK_TOKEN;

// Initialize
const web = new WebClient(token);
Initializing without a token

Alternatively, you can create a client without an token, and use it with multiple workspaces as long as you supply a token when you call a method.

const { WebClient } = require('@slack/web-api');

// Initialize a single instance for the whole app
const web = new WebClient();

// Find a token in storage (database) before making an API method call
(async () => {
  // Some fictitious database
  const token = await db.findTokenByTeam(teamId, enterpriseId)

  // Call the method
  const result = web.auth.test({ token });
})();

Call a method

The client instance has a named method for each of the public methods in the Web API. The most popular one is called chat.postMessage, and its used to send a message to a conversation. For every method, you pass arguments as properties of an options object. This helps with the readablility of your code since every argument has a name. All named methods return a Promise which resolves with the response data, or rejects with an error.

// Given some known conversation ID (representing a public channel, private channel, DM or group DM)
const conversationId = '...';

(async () => {

  // Post a message to the channel, and await the result.
  // Find more arguments and details of the response: https://api.slack.com/methods/chat.postMessage
  const result = await web.chat.postMessage({
    text: 'Hello world!',
    channel: conversationId,
  });

  // The result contains an identifier for the message, `ts`.
  console.log(`Successfully send message ${result.ts} in conversation ${conversationId}`);
})();

Hint: If you’re using an editor that supports TypeScript, even if you’re not using TypeScript to write your code, you’ll get hints for all the arguments each method supports. This helps you save time by reducing the number of times you need to pop out to a webpage to check the reference. There’s more information about using TypeScript with this package in the documentation website.

Note: Use the Block Kit Builder for a playground where you can prototype your message’s look and feel.

Using a dynamic method name

If you want to provide the method name as a string, so that you can decide which method to call dynamically, or to call a method that might not be available in your version of the client, use the WebClient.apiCall(methodName, [options]) method. The API method call above can also be written as follows:

const conversationId = '...';
(async () => {

  // Using apiCall() allows the app to call any method and to do it programmatically
  const response = await web.apiCall('chat.postMessage', {
    text: 'Hello world!',
    channel: conversationId,
  });
})();

Handle errors

Errors can happen for many reasons: maybe the token doesn’t have the proper scopes to call a method, maybe its been revoked by a user, or maybe you just used a bad argument. In these cases, the returned Promise will reject with an Error. You should catch the error and use the information it contains to decide how your app can proceed.

Each error contains a code property, which you can check against the ErrorCode export to understand the kind of error you’re dealing with. For example, when Slack responds to your app with an error, that is an ErrorCode.PlatformError. These types of errors provide Slack’s response body as the data property.

// Import ErrorCode from the package
const { WebClient, ErrorCode } = require('@slack/web-api');

(async () => {

  try {
    // This method call should fail because we're giving it a bogus user ID to lookup.
    const response = await web.users.info({ user: '...' });
  } catch (error) {
    // Check the code property, and when its a PlatformError, log the whole response.
    if (error.code === ErrorCode.PlatformError) {
      console.log(error.data);
    } else {
      // Some other error, oh no!
      console.log('Well, that was unexpected.');
    }
  }
})();
More error types

There are a few more types of errors that you might encounter, each with one of these codes:

  • ErrorCode.RequestError: A request could not be sent. A common reason for this is that your network connection is not available, or api.slack.com could not be reached. This error has an original property with more details.

  • ErrorCode.RateLimitedError: The Web API cannot fulfill the API method call because your app has made too many requests too quickly. This error has a retryAfter property with the number of seconds you should wait before trying again. See the documentation on rate limit handling to understand how the client will automatically deal with these problems for you.

  • ErrorCode.HTTPError: The HTTP response contained an unfamiliar status code. The Web API only responds with 200 (yes, even for errors) or 429 (rate limiting). If you receive this error, its likely due to a problem with a proxy, a custom TLS configuration, or a custom API URL. This error has the statusCode, statusMessage, headers, and body properties containing more details.


Pagination

Many of the Web API’s methods return lists of objects, and are known to be cursor-paginated. The result of calling these methods will contain a part of the list, or a page, and also provide you with information on how to continue to the next page on a subsequent API call. Instead of calling many times manually, the WebClient can manage getting each page, allowing you to determine when to stop, and help you process the results.

The process of retrieving multiple pages from Slack’s API can be described as asynchronous iteration, which means you’re processing items in a collection, but getting each item is an asynchronous operation. Fortunately, JavaScript has this concept built in, and in newer versions of the language there’s syntax to make it even simpler: for await...of.

(async () => {
  let result;

  // Async iteration is similar to a simple for loop.
  // Use only the first two parameters to get an async iterator.
  for await (const page of web.paginate('something.list', { name: 'value' })) {
    // You can inspect each page, find your result, and stop the loop with a `break` statement
    if (containsTheThing(page.something)) {
      result = page.something.thing;
      break;
    }
  }
});

The for await...of syntax is available in Node v10.0.0 and above. If you’re using an older version of Node, see functional iteration below.

Using functional iteration

The .paginate() method can accept up to two additional parameters. The third parameter, stopFn, is a function that is called once for each page of the result, and should return true when the app no longer needs to get another page. The fourth parameter is reducerFn, which is a function that gets called once for each page of the result, but can be used to aggregate a result. The value it returns is used to call it the next time as the accumulator. The first time it gets called, the accumulator is undefined.

(async () => {

  // The first two parameters are the method name and the options object.
  const done = await web.paginate('something.list', { name: 'value' },
    // The third is a function that receives each page and should return true when the next page isn't needed.
    (page) => { /* ... */ },
    // The fourth is a reducer function, similar to the callback parameter of Array.prototype.reduce().
    // https://developer.mozilla.org/en-US/docs/Web/JavaScript/Reference/Global_Objects/Array/Reduce
    // The accumulator is initialized to undefined.
    (accumulator, page, index) => { /* ... */ },
  );
});

The returned value is a Promise, but what it resolves to depends on whether or not you include the fourth (optional) parameter. If you don’t include it, the resolved value is always undefined. In this case, its used for control flow purposes (resuming the rest of your program), and the function in the third parameter is used to capture a result. If you do include the fourth parameter, then the resolved value is the value of the accumulator. This is a familiar pattern for people that use functional programming.


Logging

The WebClient will log interesting information to the console by default. You can use the logLevel to decide how much information, or how interesting the information needs to be, in order for it to be output. There are a few possible log levels, which you can find in the LogLevel export. By default, the value is set to LogLevel.INFO. While you’re in development, its sometimes helpful to set this to the most verbose: LogLevel.DEBUG.

// Import LogLevel from the package
const { WebClient, LogLevel } = require('@slack/web-api');

// Log level is one of the options you can set in the constructor
const web = new WebClient(token, {
  logLevel: LogLevel.DEBUG,
});

All the log levels, in order of most to least information are: DEBUG, INFO, WARN, and ERROR.

Sending log output somewhere besides the console

You can also choose to have logs sent to a custom logger using the logger option. A custom logger needs to implement specific methods (known as the Logger interface):

Method Parameters Return type
setLevel() level: LogLevel void
setName() name: string void
debug() ...msgs: any[] void
info() ...msgs: any[] void
warn() ...msgs: any[] void
error() ...msgs: any[] void

A very simple custom logger might ignore the name and level, and write all messages to a file.

const { createWriteStream } = require('fs');
const logWritable = createWriteStream('/var/my_log_file'); // Not shown: close this stream

const web = new WebClient(token, {
  // Creating a logger as a literal object. It's more likely that you'd create a class.
  logger: {
    debug(...msgs): { logWritable.write('debug: ' + JSON.stringify(msgs)); },
    info(...msgs): { logWritable.write('info: ' + JSON.stringify(msgs)); },
    warn(...msgs): { logWritable.write('warn: ' + JSON.stringify(msgs)); },
    error(...msgs): { logWritable.write('error: ' + JSON.stringify(msgs)); },
    setLevel(): { },
    setName(): { },
  },
});

Automatic retries

In production systems, you want your app to be resilient to short hiccups and temporary outages. Solving for this problem usually involves building a queuing system that handles retrying failed tasks. The WebClient comes with this queuing system out of the box, and its on by default! The client will retry a failed API method call up to 10 times, spaced out over about 30 minutes. If the request doesn’t succeed in that time, then the returned Promise will reject. You can observe each of the retries in your logs by setting the log level to DEBUG. Try running the following code with your network disconnected, and then re-connect after you see a couple of log messages:

const { WebClient, LogLevel } = require('@slack/web-api');

const web = new WebClient('bogus token');

(async () => {
  await web.auth.test();

  console.log('Done!');
})();

Shortly after re-connecting your network, you should see the Done! message. Did you notice the program doesn’t use a valid token? The client is doing something clever and helpful here. It knows the difference between an error such as not being able to reach api.slack.com and an error in the response from Slack about an invalid token. The former is something that can be resolved with a retry, so it was retried. The invalid token error means that the call isn’t going to succeed until your app does something differently, so it stops attempting retries.

You might not think 10 reties in 30 minutes is a good policy for your app. No problem, you can set the retryConfig to one that works better for you. The retryPolicies export contains a few well known options, and you can always write your own.

const { WebClient, retryPolicies } = require('@slack/web-api');

const web = new WebClient(token, {
  retryConfig: retryPolicies.fiveRetriesInFiveMinutes,
});

Here are some other values that you might want to use for retryConfig:

retryConfig Description
retryPolicies.tenRetriesInAboutThirtyMinutes (default)
retryPolicies.fiveRetriesInFiveMinutes Five attempts in five minutes
retryPolicies.rapidRetryPolicy Used to keep tests running fast
{ retries: 0 } No retries (other options)

Note: If an API call results in a rate limit being exceeded, you might still notice the client automatically retrying the API call. If you’d like to opt out of that behavior, set the rejectRateLimitedCalls option to true.


Upload a file

A couple methods, files.upload and users.setPhoto, allow you to upload a file over the API. In Node, there are a few ways you might be dealing with files, or more generally, binary data. When you have the whole file in memory (like when you’ve just generated or processed an image), then in Node you’d have a Buffer that contains that binary data. Or, when you are reading the file from disk or a network (like when you have a path to file name), then you’d typically have a ReadableStream. The client can handle both of these binary data types for you, and it looks like any other API call.

The following example shows how you can use files.upload to upload a file that is read from disk (as a ReadableStream).

const { createReadStream } = require('fs');
const { WebClient } = require('@slack/web-api');

const token = process.env.SLACK_TOKEN;
const web = new WebClient(token);

// A file name is required for the upload
const filename = 'test_file.csv';

(async () => {
  // Just use the `file` argument as the documentation suggests
  // See: https://api.slack.com/methods/files.upload
  const result = await web.files.upload({
    filename,
    // You can use a ReadableStream or a Buffer for the file option
    // This file is located in the current directory (`process.pwd()`), so the relative path resolves
    file: createReadStream(`./${fileName}`),
  })

  // `res` contains information about the uploaded file
  console.log('File uploaded: ', result.file.id);
})();

In the example above, you could also use a Buffer object as the value for the file property of the options object.


Proxy requests with a custom agent

The client allows you to customize the HTTP Agent used to create the connection to Slack. Using this option is the best way to make all requests from your app through a proxy, which is a common requirement in many corporate settings.

In order to create an Agent from some proxy information (such as a host, port, username, and password), you can use one of many npm packages. We recommend https-proxy-agent. Start by installing this package and saving it to your package.json.

$ npm install https-proxy-agent

Import the HttpsProxyAgent class, and create an instance that can be used as the agent option of the WebClient.

const { WebClient } = require('@slack/web-api');
const { HttpsProxyAgent } = require('https-proxy-agent');
const token = process.env.SLACK_TOKEN;

// One of the ways you can configure HttpsProxyAgent is using a simple string.
// See: https://github.com/TooTallNate/node-https-proxy-agent for more options
const proxy = new HttpsProxyAgent(process.env.http_proxy || 'http://168.63.76.32:3128');

const web = new WebClient(token, { agent: proxy });

// All API method calls will now go through the proxy

Rate limits

When your app calls API methods too frequently, Slack will politely ask (by returning an error) the app to slow down, and also let your app know how many seconds later it should try again. This is called rate limiting and the WebClient handles it for your app with grace. The client will understand these rate limiting errors, wait the appropriate amount of time, and then retry the request without any changes in your code. The Promise returned only resolves when Slack has given your app a real response.

It’s a good idea to know when you’re bumping up against these limits, so that you might be able to change the behavior of your app to hit them less often. Your users would surely appreciate getting things done without the delay. Each time a rate limit related error occurs, the WebClient instance emits an event: WebClientEvent.RATE_LIMITED. We recommend that you use the event to inform users when something might take longer than expected, or just log it for later.

const { WebClient, WebClientEvent } = require('@slack/web-api');
const token = process.env.SLACK_TOKEN;

const web = new WebClient(token);

web.on(WebClientEvent.RATE_LIMITED, (numSeconds) => {
  console.log(`A rate-limiting error occurred and the app is going to retry in ${numSeconds} seconds.`);
});

You might not want to the WebClient to handle rate limits in this way. Perhaps the operation was time sensitive, and it won’t be useful by the time Slack is ready for another request. Or, you have a more sophisticated approach. In these cases, you can set the rejectRateLimitedCalls option on the client to true. Once you set this option, method calls can fail with rate limiting related errors. These errors have a code property set to ErrorCode.RateLimitedError. See error handling for more details.

const { WebClient } = require('@slack/web-api');
const token = process.env.SLACK_TOKEN;

// Setting the rejectRateLimitedCalls option to true turns off automatic retries
const web = new WebClient(token, { rejectRateLimitedCalls: true });

Request concurrency

Each of the API method calls the client starts are happening concurrently, or at the same time. If your app tries to perform a lot of method calls, let’s say 100 of them, at the same time, each one of them would be competing for the same network resources (such as bandwidth). By competing, they might negatively affect the performance of all the rest, and therefore negatively affect the performance of your app. This is one of the reasons why the WebClient limits the concurrency of requests by default to ten, which means it keeps track of how many requests are waiting, and only starts an eleventh request when one of them completes. The exact number of requests the client allows at the same time can be set using the maxRequestConcurrency option.

const { WebClient } = require('@slack/web-api');
const token = process.env.SLACK_TOKEN;

// Limit the number of concurrent requests to 5
const web = new WebClient(token, { maxRequestConcurrency: 5 });

The lower you set the maxRequestConcurrency, the less parallelism you’ll have in your app. Imagine setting the concurrency to 1. Each of the method calls would have to wait for the previous method call to complete before it can even be started. This could slow down your app significantly. So its best not to set this number too low.

Another reason, besides competing for resources, that you might limit the request concurrency is to minimize the amount of state in your app. Each request that hasn’t completed is in some ways a piece of state that hasn’t yet been stored anywhere except the memory of your program. In the scenario where you had 100 method calls waiting, and your program unexpectedly crashes, you’ve lost information about 100 different things going on in the app. But by limiting the concurrency to a smaller number, you can minimize this risk. So its best not to set this number too high.


Custom TLS configuration

Each connection to Slack starts with a handshake that allows your app to trust that it is actually Slack you are connecting to. The system for establishing this trust is called TLS. In order for TLS to work, the host running your app keeps a list of trusted certificate authorities, that it can use to verify a signature Slack produces. You don’t usually see this list, its usually a part of the operating system you’re running on. In very special cases, like certain testing techniques, you might want to send a request to another party that doesn’t have a valid TLS signature that your certificate authority would verify. In these cases, you can provide alternative TLS settings, in order to change how the operating system might determine whether the signature is valid. You can use the tls option to describe the settings you want (these settings are the most common and useful from the standard Node API).

const { WebClient } = require('@slack/web-api');
const { readFileSync } = require('fs');
const token = process.env.SLACK_TOKEN;

// Load a custom certificate authority from a file in the current directory
const ca = readFileSync('./ca.crt');

// Initialize a client with the custom certificate authority
const web = new WebClient(token, { tls: { ca } });
tls property Description
ca Optionally override the trusted CA certificates. Any string or Buffer can contain multiple PEM CAs concatenated together.
key Private keys in PEM format. PEM allows the option of private keys being encrypted. Encrypted keys will be decrypted with passphrase.
cert Cert chains in PEM format. One cert chain should be provided per private key.
pfx PFX or PKCS12 encoded private key and certificate chain. pfx is an alternative to providing key and cert individually. PFX is usually encrypted, if it is, passphrase will be used to decrypt it.
passphrase Shared passphrase used for a single private key and/or a PFX.

Custom API URL

The URLs for method calls to Slack’s Web API always begin with https://slack.com/api/. In very special cases, such as certain testing techniques, you might want to send these requests to a different URL. The slackApiUrl option allows you to replace this prefix with another.

const { WebClient } = require('@slack/web-api');
const token = process.env.SLACK_TOKEN;

const options = {};

// In a testing environment, configure the client to send requests to a mock server
if (process.env.NODE_ENV === 'test') {
  options.slackApiUrl = 'http://localhost:8888/api/';
}

// Initialize a client using the configuration
const web = new WebClient(token, options);

Exchange an OAuth grant for a token

There’s one method in the Slack Web API that doesn’t requires a token, because its the method that gets a token! This method is called oauth.access. It’s used as part of the OAuth 2.0 process that users initiate when installing your app into a workspace. In the last step of this process, your app has received an authorization grant called code which it needs to exchange for an access token (token). You can use an instance of the WebClient that has no token to easily complete this exchange.

const { WebClient } = require('@slack/web-api');

// App credentials found in the Basic Information section of the app configuration
const clientId = process.env.SLACK_CLIENT_ID;
const clientSecret = process.env.SLACK_CLIENT_SECRET;

// Not shown: received an authorization grant called `code`.
(async () => {
  // Create a client instance just to make this single call, and use it for the exchange
  const result = await (new WebClient()).oauth.access({
    client_id: clientId,
    client_secret: clientSecret,
    code
  });

  // It's now a good idea to save the access token to your database
  // Some fictitious database
  await db.createAppInstallation(result.team_id, result.enterprise_id, result.access_token, result.bot);
})();

Note: If you’re looking for a more complete solution that handles more of the OAuth process for your app, take a look at the @aoberoi/passport-slack Passport Strategy.