feat: accept chunks as arguments to chat.{start,append,stop}Stream methods #2467

srtaalej · 2026-01-11T21:52:56Z

Summary

This PR introduces the chunks argument to the following methods:

chat.startStream
chat.appendStream
chat.stopStream

⚠️ This feature is experimental at the moment!

Code Snippet

const stream = new ChatStreamer(client, client.logger, {
      channel: CHANNEL_ID,
      thread_ts: threadTs,
      recipient_team_id: 'T09EX3FS2J3',
      recipient_user_id: 'U09G4FG3TRN',
    }, { buffer_size: 1 });

    await stream.append({
      chunks: [
        {
          type: 'markdown_text',
          text: '**Hello!** I am starting to process your request...\n\n'
        }]
    });
   
    await stream.append({
      chunks: [
        {
          type: 'plan_update',
          title: ''\nProcessing tasks...\n,
        },
        {
          type: 'task_update',
          id: 'task-1',
          title: 'Fetching data from API',
          status: 'complete',
          output: 'Successfully retrieved 42 records',
        },
        {
          type: 'task_update',
          id: 'task-2',
          title: 'Processing results',
          status: 'pending',
        },
      ],
    });
    
    await stream.append({
      chunks: [
        {
          type: 'task_update',
          id: 'task-2',
          title: 'Processing results',
          status: 'complete',
          output: 'Analysis complete!',
          sources: [
            {
              type: 'url',
              url: 'https://example.com/docs',
              text: 'Documentation',
            },
            {
              type: 'url',
              url: 'https://example.com/api',
              text: 'API Reference',
              icon_url: 'https://example.com/favicon.ico',
            },
          ],
        },
      ],
    });
   
     await stream.append({
      chunks: [
        {
          type: 'markdown_text',
          text: 'Hi Ale!\n\n',
        },
        {
          type: 'markdown_text',
          text: 'Here is what I found:\n\n',
        },
        {
          type: 'markdown_text',
          text: '- Item 1: *Important information*\n',
        },
        {
          type: 'markdown_text',
          text: '- Item 2: `Some code snippet`\n',
        },
      ],
    });

   
    await stream.stop({
      chunks: [
        {
          type: 'markdown_text',
          text: '\n\n---\n\n✅ **All tasks completed successfully!**\n',
        },
        {
          type: 'markdown_text',
          text: 'Thank you for using the streaming API with chunks.',
        },
      ],
    });

Requirements (place an `x` in each `[ ]`)

I've read and understood the Contributing Guidelines and have done my best effort to follow them.
I've read and agree to the Code of Conduct.

codecov · 2026-01-11T21:53:44Z

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 93.09%. Comparing base (b44cee1) to head (1486d3f).
⚠️ Report is 1 commits behind head on feat-ai-apps-thinking-steps.
✅ All tests successful. No failed tests found.

Additional details and impacted files

@@                     Coverage Diff                      @@
##           feat-ai-apps-thinking-steps    #2467   +/-   ##
============================================================
  Coverage                        93.09%   93.09%           
============================================================
  Files                               40       40           
  Lines                            11240    11240           
  Branches                           713      713           
============================================================
  Hits                             10464    10464           
  Misses                             764      764           
  Partials                            12       12

Flag	Coverage Δ
cli-hooks	`95.23% <ø> (ø)`
cli-test	`94.79% <ø> (ø)`
webhook	`96.66% <ø> (ø)`

Flags with carried forward coverage won't be shown. Click here to find out more.

🚀 New features to boost your workflow:

📦 JS Bundle Analysis: Save yourself from yourself by tracking and limiting bundle sizes in JS merges.

…thods

zimeg

@srtaalej Woooo! This'll be nice to land soon but I'm leaving a few comments on changes perhaps to make beforehand.

The tests are most confusing to me since I think all changes are contained in this PR... Am looking into this separate but do let me know if questions on these comments appear please!

packages/types/src/index.ts

packages/types/src/chunk.ts

zimeg · 2026-01-21T17:46:00Z

packages/types/src/chunk.ts

+  if(type === 'plan_update') {
+    if (typeof chunkObj.title === 'string') {
+      return chunkObj as unknown as PlanUpdateChunk;
+    }
+    console.warn('Invalid PlanUpdateChunk (missing title property)', chunk);
+    return null;
+  }


⚠️ suggestion: It might be preferable to let the API return an error for invalid chunks since these limitations might change in the future?

packages/types/src/chunk.ts

packages/web-api/test/types/methods/chat.test-d.ts

packages/types/src/chunk.ts

Co-authored-by: Eden Zimbelman <eden.zimbelman@salesforce.com>

mwbrooks

🙌🏻 This is looking great @srtaalej! Thanks for aligning so closely with the Python SDK implement as well.

🧪 I've double-checked that the params are correct and thanks for adding the detailed tests to chat.test-d.ts. 🙇🏻

Here is my modification to our sample app's listeners/assistant/message.ts method to test the chat.startStream, chat.appendStream, and chat.stopStream with markdown_text, and task_update. I don't think I can test plan_update until the PR that adds task_display_mode.

listeners/assistant/message.ts

import { DEFAULT_SYSTEM_CONTENT, openai } from '../../ai/index.js';
import { feedbackBlock } from '../views/feedback_block.js';

const delay = (ms) => new Promise(resolve => setTimeout(resolve, ms));

/**
 * The `message` event is sent when the user direct messages the app in a DM or Assistant container.
 *
 * @param {Object} params
 * @param {import("@slack/web-api").WebClient} params.client - Slack web client.
 * @param {import("@slack/bolt").Context} params.context - Event context.
 * @param {import("@slack/logger").Logger} params.logger - Logger instance.
 * @param {import("@slack/types").MessageEvent} params.message - The incoming message.
 * @param {Function} params.getThreadContext - Function to get thread context.
 * @param {import("@slack/bolt").SayFn} params.say - Function to send messages.
 * @param {Function} params.setTitle - Function to set assistant thread title.
 * @param {Function} params.setStatus - Function to set assistant status.
 *
 * @see {@link https://docs.slack.dev/reference/events/message}
 */
export const message = async ({ client, context, logger, message, getThreadContext, say, setTitle, setStatus }) => {
  /**
   * Messages sent to the Assistant can have a specific message subtype.
   *
   * Here we check that the message has "text" and was sent to a thread to
   * skip unexpected message subtypes.
   *
   * @see {@link https://docs.slack.dev/reference/events/message#subtypes}
   */
  if (!('text' in message) || !('thread_ts' in message) || !message.text || !message.thread_ts) {
    return;
  }
  const { channel, thread_ts } = message;
  const { userId, teamId } = context;

  try {
    /**
     * Set the title of the Assistant thread to capture the initial topic/question
     * as a way to facilitate future reference by the user.
     *
     * @see {@link https://docs.slack.dev/reference/methods/assistant.threads.setTitle}
     */
    await setTitle(message.text);

    /**
     * Set the status of the Assistant to give the appearance of active processing.
     *
     * @see {@link https://docs.slack.dev/reference/methods/assistant.threads.setStatus}
     */
    await setStatus({
      status: 'thinking...',
      loading_messages: [
        'Teaching the hamsters to type faster…',
        'Untangling the internet cables…',
        'Consulting the office goldfish…',
        'Polishing up the response just for you…',
        'Convincing the AI to stop overthinking…',
      ],
    });

    const streamResponse = await client.chat.startStream({
      channel: channel,
      thread_ts: thread_ts,
      recipient_team_id: teamId,
      recipient_user_id: userId,
      chunks: [{
        type: 'markdown_text',
        text: '**Hello!** I am starting to process your request...\n\n',
      }, {
        type: 'task_update',
        id: '12',
        title: 'counting bytes...',
        status: 'in_progress',
      }],
    });

    await delay(4000);

    await client.chat.appendStream({
      channel: channel,
      ts: streamResponse.ts,
      // markdown_text: '',
      chunks: [{
        type: 'task_update',
        id: '12',
        title: 'adding numbers...',
        status: 'in_progress',
        details: 'sums have increased',
      }],
    });

    await delay(4000);

    await client.chat.stopStream({
      channel: channel,
      ts: streamResponse.ts,
      chunks: [{
        type: 'task_update',
        id: '12',
        title: 'solved equation!',
        status: 'complete',
        sources: [{
          type: 'url',
          url: 'https://oeis.org',
          text: 'The On-Line Encyclopedia of Integer Sequences (OEIS)',
        }],
      }, {
        type: 'markdown_text',
        text: 'that computes.',
      }],
    });
  } catch (e) {
    logger.error(e);

    // Send message to advise user and clear processing status if a failure occurs
    await say({ text: `Sorry, something went wrong! ${e}` });
  }
};

📝 I left a suggestion to remove the parseChunk and npm dependency. After that's done, I think we can ✅ and merge!

mwbrooks · 2026-01-23T00:08:25Z

packages/types/src/chunk.ts

+/**
+ * Parse a chunk object and return the appropriate typed chunk.
+ * Returns null if the chunk is invalid or unknown.
+ */
+export function parseChunk(chunk: unknown): AnyChunk | null {


suggestion: This is a nice implementation, but I don't think we need it for the @slack/types package. This parsing was necessary for Python, but in TypeScript I think the AnyChunk union that you created will serve well! :two-cents:

ahh okay thank you for the clarification 🙌

mwbrooks · 2026-01-23T00:09:38Z

packages/types/package.json

+  "dependencies": {
+    "@slack/logger": "^4.0.0"
+  },


suggestion: If we remove the parseChunk method, then we can also remove this dependency 🎉

mwbrooks · 2026-01-23T00:28:33Z

packages/web-api/test/types/methods/chat.test-d.ts

 expectError(
  web.chat.appendStream({
    channel: 'C1234', // missing_markdown_text
    ts: '1234.56',
  }),
 );


suggestion: @srtaalej I manually linked the new @slack/types into this package and ran npm test. It looks like this test now no longer errors because markdown_text is now optional. However, I'd still expect it to error when markdown_text and chunks are both missing. 🤔 @zimeg Do you think this error should come from the Slack API or should our SDK check that at least one of these two params exist?

> @slack/web-api@7.13.0 test:types > tsd test/types/methods/chat.test-d.ts:21:0 ✖ 21:0 Expected an error, but found none. 1 error

👁️‍🗨️ @mwbrooks Nice catch! I'd lean toward documenting that either markdown_text or chunks is required and letting the API error instead of leaning on typescript options.

🗣️ ramble: That might be personal preference, but IMHO these options don't catch runtime nuance that the call to an API would anyways. An empty markdown_text string IIRC causes a no_text error from API but the required markdown_text appears alright.

mwbrooks

✅ Thanks for the update @srtaalej! 🙇🏻

🧪 Things look good on my side and tests pass locally when I npm link the @slack/types package in the package/web-api.

Let's merge this and keep moving forward! 🚀

feat: accept chunks as arguments to chat.{start,append,stop}Stream me…

e3d7b9b

…thods

srtaalej force-pushed the ale-chunks branch from 225f47f to e3d7b9b Compare January 11, 2026 21:57

srtaalej self-assigned this Jan 11, 2026

srtaalej added the enhancement M-T: A feature request for new functionality label Jan 11, 2026

build: update slack/types version in package json

2601199

srtaalej changed the base branch from main to feat-ai-apps-thinking-steps January 12, 2026 19:51

srtaalej marked this pull request as ready for review January 15, 2026 22:48

srtaalej requested review from a team as code owners January 15, 2026 22:48

fix: make markdown text optional and add missing chunk type

160cfb7

srtaalej force-pushed the ale-chunks branch from 749493a to 160cfb7 Compare January 16, 2026 16:52

tests: add missing chunk type to tests

81140cc

srtaalej added the semver:minor label Jan 20, 2026

srtaalej requested a review from lukegalbraithrussell January 20, 2026 18:56

srtaalej assigned misscoded and unassigned misscoded Jan 20, 2026

srtaalej requested a review from misscoded January 20, 2026 18:57

zimeg reviewed Jan 21, 2026

View reviewed changes

zimeg added pkg:web-api applies to `@slack/web-api` pkg:types applies to `@slack/types` labels Jan 21, 2026

zimeg reviewed Jan 21, 2026

View reviewed changes

packages/types/src/chunk.ts Outdated Show resolved Hide resolved

srtaalej mentioned this pull request Jan 21, 2026

feat: showcase thinking steps slack-samples/bolt-js-assistant-template#98

Open

5 tasks

srtaalej and others added 4 commits January 22, 2026 13:27

build: use @slack/logger for debugging parsechunks

dbca9c9

type narrowing for parseChunks

1d480f6

Update packages/web-api/test/types/methods/chat.test-d.ts

e6ece09

Co-authored-by: Eden Zimbelman <eden.zimbelman@salesforce.com>

Update packages/web-api/test/types/methods/chat.test-d.ts

77d3986

Co-authored-by: Eden Zimbelman <eden.zimbelman@salesforce.com>

mwbrooks reviewed Jan 23, 2026

View reviewed changes

remove parsechunks

c6c52c8

merge remote changes into local

1486d3f

mwbrooks approved these changes Jan 23, 2026

View reviewed changes

srtaalej merged commit e410413 into feat-ai-apps-thinking-steps Jan 23, 2026
39 of 57 checks passed

srtaalej deleted the ale-chunks branch January 23, 2026 18:45

feat: accept chunks as arguments to chat.{start,append,stop}Stream methods #2467

feat: accept chunks as arguments to chat.{start,append,stop}Stream methods #2467

Conversation

srtaalej commented Jan 11, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Summary

Requirements (place an x in each [ ])

Uh oh!

codecov bot commented Jan 11, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Codecov Report

Uh oh!

zimeg left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

zimeg Jan 21, 2026

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

mwbrooks left a comment

Choose a reason for hiding this comment

Uh oh!

mwbrooks Jan 23, 2026

Choose a reason for hiding this comment

Uh oh!

srtaalej Jan 23, 2026

Choose a reason for hiding this comment

Uh oh!

mwbrooks Jan 23, 2026

Choose a reason for hiding this comment

Uh oh!

mwbrooks Jan 23, 2026

Choose a reason for hiding this comment

Uh oh!

zimeg Jan 23, 2026

Choose a reason for hiding this comment

Uh oh!

mwbrooks left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

5 participants

srtaalej commented Jan 11, 2026 •

edited

Loading

Requirements (place an `x` in each `[ ]`)

codecov bot commented Jan 11, 2026 •

edited

Loading