feat: showcase thinking steps by srtaalej · Pull Request #98 · slack-samples/bolt-js-assistant-template

srtaalej · 2026-01-21T22:11:25Z

Type of change

New feature
Bug fix
Documentation

Summary

This PR showcases "chunks" in chat streams. Related: slackapi/node-slack-sdk#2467

⚠️ the chunks type is still under construction resulting in the linter error below 😿

Demo videos 📹

dice-demo.mov

wonder-demo.mov

Requirements

I have ensured the changes I am contributing align with existing patterns and have tested and linted my code
I've read and agree to the Code of Conduct

srtaalej · 2026-01-21T22:14:39Z

README.md

@@ -6,7 +6,7 @@ Models from [OpenAI](https://openai.com) are used and can be customized for prom

 ## Setup


note - still deciding how to restructure README.md to make setup easier. I think the most obvious point of confusion is deciding whether to use CLI or terminal to setup, i think a toggle like
Details
might make things simpler 🤔 ?

@srtaalej Thanks for calling this out! 📚 ✨ I agree it's a confusing and hope we can encourage the CLI most-

IIRC @lukegalbraithrussell had thoughts on related changes as well and we should match such in slack-samples/bolt-python-assistant-template#42 too!

zimeg

@srtaalej Nice - I'm leaving a few quick comments now but am looking forward to testing this! 📺 ✨

listeners/assistant/message.js

zimeg · 2026-01-22T05:14:03Z

listeners/assistant/message.js

-        input: `System: ${DEFAULT_SYSTEM_CONTENT}\n\nUser: ${llmPrompt}`,
-        stream: true,
+    // This first example shows a generated text response for the provided prompt
+    if (message.text !== 'Wonder a few deep thoughts.') {


📝 note: This != and the cases contained were switched in the final commits of slack-samples/bolt-python-assistant-template#37 which we might want to match?

agent/llm_caller.js

zimeg · 2026-01-22T05:20:45Z

README.md

@@ -6,7 +6,7 @@ Models from [OpenAI](https://openai.com) are used and can be customized for prom

 ## Setup


@srtaalej Thanks for calling this out! 📚 ✨ I agree it's a confusing and hope we can encourage the CLI most-

IIRC @lukegalbraithrussell had thoughts on related changes as well and we should match such in slack-samples/bolt-python-assistant-template#42 too!

zimeg

@srtaalej Sweet I'm liking the results of a roll I prompted 🏆

I'm requesting a few changes in hopes of focusing most on tool calls with this example, and also want to showcase the buffer in text streaming 🤖

Tests related to type checking might need a small change too, but once we have CI passing with a prerelease I think we should look toward merging this!

zimeg · 2026-01-26T21:04:08Z

agent/tools/dice.js

+/**
+ * Tool definition for OpenAI API
+ *
+ * @see {@link https://platform.openai.com/docs/guides/function-calling}
+ */
+export const rollDiceDefinition = {


🪬 note: Using jsdoc to match the expected type of the openai.responses.create arguments might be required since we're exporting from this package? I'm unsure what that type might be, but am hoping it'd address the error above!

listeners/assistant/assistant_thread_started.js

agent/llm_caller.js

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

mwbrooks

🙌🏻 Thanks for getting the thinking steps demo together! It's looking very closely aligned to the Python demo 👌🏻

✏️ I've left a few suggestions and I'll begin manual testing next!

agent/llm_caller.js

package.json

listeners/assistant/message.js

listeners/assistant/assistant_thread_started.js

listeners/assistant/message.js

mwbrooks · 2026-01-27T22:35:47Z

listeners/assistant/message.js

+    const { channel, thread_ts } = message;
+    const { userId, teamId } = context;
+
+    // The first example shows detailed thinking steps similar to tool calls


question: Should this comment be swapped with the else clause? The if seems to handle the scripted "Wonder a thought" example while the else handles the roll-the-dice tool call.

@mwbrooks We should reword this to be more clear in both samples IMHO, but the order I think is correct. I'm curious of other changes too, but might suggest:

The first example shows a message with thinking steps that has different chunks to construct and update a plan alongside text outputs.

mwbrooks · 2026-01-27T22:36:53Z

listeners/assistant/message.js

-        model: 'gpt-4o-mini',
-        input: `System: ${DEFAULT_SYSTEM_CONTENT}\n\nUser: ${llmPrompt}`,
-        stream: true,
+      await streamer.stop({


suggestion: I think we should add the feedback blocks to this .stop (and add it to Python if it's missing).

mwbrooks · 2026-01-27T22:38:07Z

agent/llm_caller.js

+ * @see {@link https://platform.openai.com/docs/guides/streaming-responses}
+ * @see {@link https://platform.openai.com/docs/guides/function-calling}
+ */
+export async function callLlm(streamer, prompts) {


suggestion: Take it or leave it, but I feel callLlm is a funny syntax in JavaScript. Instead, I'd expect callLLM. .cc @zimeg @srtaalej

🐣 note: I'm open to that! But am unsure of best practices for this. As long as it's not the following-

callllm

mwbrooks · 2026-01-27T22:39:31Z

manifest.json

-        "assistant_thread_started",
-        "message.im"
-      ]
+      "bot_events": ["app_mention", "assistant_thread_context_changed", "assistant_thread_started", "message.im"]


suggestion: I'd suggest putting each item of the array on a newline to match the JSON file format.

mwbrooks

🧪 Manually testing the sample goes works g-g-g-g-g-great! 👏🏻

zimeg

💡 Praises to these changes so far! Thanks! ✨

Responses are generating well for me but I left a few more comments with changes for CI checks. Some of these might find a test with this command too:

$ npm run check

But I realize that's not found with a proper "test" script 🧪

The comments @mwbrooks shared are all solid finds too! Once these are finalized let's mirror similar updates to Python implementation 🙏

zimeg · 2026-01-28T06:06:11Z

listeners/assistant/message.js

+    const { channel, thread_ts } = message;
+    const { userId, teamId } = context;
+
+    // The first example shows detailed thinking steps similar to tool calls


@mwbrooks We should reword this to be more clear in both samples IMHO, but the order I think is correct. I'm curious of other changes too, but might suggest:

The first example shows a message with thinking steps that has different chunks to construct and update a plan alongside text outputs.

agent/tools/dice.js

zimeg · 2026-01-28T06:10:55Z

agent/llm_caller.js

+ * @see {@link https://platform.openai.com/docs/guides/streaming-responses}
+ * @see {@link https://platform.openai.com/docs/guides/function-calling}
+ */
+export async function callLlm(streamer, prompts) {


🐣 note: I'm open to that! But am unsure of best practices for this. As long as it's not the following-

callllm

agent/llm_caller.js

listeners/assistant/message.js

zimeg · 2026-01-28T06:20:32Z

agent/llm-caller.js

+// OpenAI LLM client
+export const openai = new OpenAI({
+  apiKey: process.env.OPENAI_API_KEY,
+});


🌚 question: Is the openai client still used outside of this package, or could we remove the export here?

good catch! it is not

Co-authored-by: Michael Brooks <mbrooks@slack-corp.com>

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

… message.js example 1

srtaalej · 2026-01-28T18:43:40Z

TY for your reviews @zimeg @mwbrooks ❤️ your suggestions have been committed and this PR is ready for re-review 😸

mwbrooks

✅ Thanks a bunch for addressing all of our feedback @srtaalej!

🧪 Manual testing works great on my side! 🕹️

📝 I've left one final suggestion to rename callLLm to callLLM (or back to callLlm). Once you've update that, I think we're good to merge it!

agent/llm_caller.js

zimeg

🗣️ Nice! I'm leaving a few more comments of polish but will approve this for now.

Let's make sure our checks are passing with development builds linked before merging this. One comment I left might require an upstream change beforehand, but I'm so glad this is caught in this PR!

Some conventions to file casing might be nice to update here as well, but this can be saved for follow up too 👾

agent/llm-caller.js

listeners/assistant/message.js

manifest.json

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

srtaalej · 2026-01-29T16:48:33Z

thank you @zimeg for the ChatStreamer export pr! i'll update here once we get that merged ⭐ ⭐ ⭐

mwbrooks · 2026-01-29T18:09:41Z

thank you @zimeg for the ChatStreamer export pr! i'll update here once we get that merged ⭐ ⭐ ⭐

@srtaalej I've merged slackapi/node-slack-sdk#2481 so the ChatStreamer export should now be available!

srtaalej · 2026-01-29T19:03:53Z

⚠️ Note: CI will continue to fail until node feat-ai-app-thinking-steps branch is published. To run locally,

git pull origin feat-ai-apps-thinking-steps
cd web-api
npm run build
npm link
cd bolt-js-assistant-template
npm link @slack/web-api

feat: showcase thinking steps

58a717b

srtaalej self-assigned this Jan 21, 2026

srtaalej requested a review from a team as a code owner January 21, 2026 22:11

srtaalej added the enhancement New feature or request label Jan 21, 2026

srtaalej requested review from lukegalbraithrussell, misscoded, mwbrooks and zimeg January 21, 2026 22:11

srtaalej commented Jan 21, 2026

View reviewed changes

srtaalej marked this pull request as draft January 21, 2026 22:15

zimeg reviewed Jan 22, 2026

View reviewed changes

srtaalej marked this pull request as ready for review January 23, 2026 21:44

zimeg requested changes Jan 26, 2026

View reviewed changes

srtaalej marked this pull request as draft January 27, 2026 16:43

srtaalej and others added 2 commits January 27, 2026 11:44

Update agent/llm_caller.js

e8bca5e

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

fix: reviewer suggestions

25d7fef

srtaalej marked this pull request as ready for review January 27, 2026 18:45

zimeg self-requested a review January 27, 2026 18:46

srtaalej and others added 2 commits January 27, 2026 16:53

docs: experimental change to README.md highlighting app install options

9bfcde4

Merge branch 'feat-ai-apps-thinking-steps' into ale-feat-chunks

378de85

mwbrooks reviewed Jan 27, 2026

View reviewed changes

mwbrooks reviewed Jan 28, 2026

View reviewed changes

zimeg reviewed Jan 28, 2026

View reviewed changes

srtaalej and others added 7 commits January 28, 2026 11:14

Update agent/llm_caller.js

4e2fadc

Co-authored-by: Michael Brooks <mbrooks@slack-corp.com>

Update listeners/assistant/assistant_thread_started.js

fd45e19

Co-authored-by: Michael Brooks <mbrooks@slack-corp.com>

Update listeners/assistant/message.js

23ac1e3

Co-authored-by: Michael Brooks <mbrooks@slack-corp.com>

Update package.json

7b4b845

Co-authored-by: Michael Brooks <mbrooks@slack-corp.com>

Update agent/llm_caller.js

991f182

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

Update agent/tools/dice.js

8a40c6d

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

fix: change callLLM func name, display feeback block and setStatus in…

913c824

… message.js example 1

srtaalej requested a review from zimeg January 28, 2026 18:42

mwbrooks approved these changes Jan 28, 2026

View reviewed changes

agent/llm_caller.js Outdated Show resolved Hide resolved

zimeg approved these changes Jan 29, 2026

View reviewed changes

agent/llm-caller.js Show resolved Hide resolved

agent/llm-caller.js Show resolved Hide resolved

listeners/assistant/message.js Show resolved Hide resolved

manifest.json Outdated Show resolved Hide resolved

srtaalej and others added 2 commits January 29, 2026 11:19

Update manifest.json

5eef6c3

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

update names in agent/

fece0bc

srtaalej merged commit 038d316 into feat-ai-apps-thinking-steps Jan 29, 2026
1 of 4 checks passed

srtaalej deleted the ale-feat-chunks branch January 29, 2026 19:04

zimeg mentioned this pull request Jan 30, 2026

feat: include status loading messages and more feedback blocks slack-samples/bolt-python-assistant-template#45

Merged

3 tasks

		@@ -6,7 +6,7 @@ Models from [OpenAI](https://openai.com) are used and can be customized for prom

		## Setup

Conversation

srtaalej commented Jan 21, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Type of change

Summary

Requirements

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

zimeg left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Choose a reason for hiding this comment

Uh oh!

zimeg left a comment

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

mwbrooks left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

mwbrooks left a comment

Choose a reason for hiding this comment

Uh oh!

zimeg left a comment

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

srtaalej commented Jan 28, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

mwbrooks left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

zimeg left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

srtaalej commented Jan 29, 2026

Uh oh!

srtaalej commented Jan 21, 2026 •

edited

Loading

srtaalej commented Jan 28, 2026 •

edited

Loading