Overview
Builds production-ready Slack bots and integrations using the Bolt framework. Handles app setup, event subscriptions, slash commands, interactive components (buttons, modals, dropdowns), message formatting with Block Kit, and deployment. Covers both simple webhook-based notifications and full conversational bots.
Instructions
1. App Setup
When creating a new Slack bot:
- Guide the user through https://api.slack.com/apps → Create New App
- Recommend "From a manifest" for faster setup
- Generate the app manifest YAML based on required scopes and features
- Set up the following based on needs:
- Bot Token Scopes:
chat:write,commands,app_mentions:read,channels:history(formessage.channels),im:history(formessage.im),users:read - Event Subscriptions:
message.channels,app_mention,message.im(in Socket Mode no request URL is needed) - Interactivity: Enable for buttons, modals, dropdowns
- Slash Commands: Register command names and descriptions
- Bot Token Scopes:
2. Project Scaffolding
Use Bolt.js (Node) or Bolt for Python:
# Node.js (run with: node --env-file=.env app.js)
npm init -y && npm install @slack/bolt
# In CI and production, commit package-lock.json and install with `npm ci`,
# which installs the exact, integrity-checked versions from the lockfile.
# Python
pip install slack-bolt
Python equivalent of the Socket Mode setup (tokens from the environment):
import os
from slack_bolt import App
from slack_bolt.adapter.socket_mode import SocketModeHandler
app = App(token=os.environ["SLACK_BOT_TOKEN"])
@app.command("/deploy")
def deploy(ack, respond, command):
ack() # within 3 seconds
respond(f"Deploying {command['text']} to staging...")
if __name__ == "__main__":
SocketModeHandler(app, os.environ["SLACK_APP_TOKEN"]).start()
Standard project structure:
slack-bot/
├── app.js # Main entry, Bolt app initialization
├── listeners/
│ ├── commands.js # Slash command handlers
│ ├── events.js # Event handlers (messages, mentions)
│ ├── actions.js # Interactive component handlers
│ └── views.js # Modal submission handlers
├── services/ # Business logic
├── blocks/ # Block Kit message templates
├── .env # SLACK_BOT_TOKEN (xoxb-), SLACK_SIGNING_SECRET, SLACK_APP_TOKEN (xapp-); gitignored
└── package.json
3. Core Patterns
Sending messages:
await client.chat.postMessage({
channel: channelId,
blocks: [/* Block Kit blocks */],
text: "Fallback text for notifications"
});
Slash commands:
app.command('/deploy', async ({ command, ack, respond }) => {
await ack();
await respond({
blocks: buildDeployConfirmation(command.text),
response_type: 'ephemeral'
});
});
Interactive modals:
app.action('button_click', async ({ body, ack, client }) => {
await ack();
await client.views.open({
trigger_id: body.trigger_id,
view: buildModalView()
});
});
Event handling:
app.event('app_mention', async ({ event, say }) => {
await say(`Hey <@${event.user}>! How can I help?`);
});
4. Block Kit Messages
Always use Block Kit for rich messages. Key block types:
section— text with optional accessory (button, image, overflow)actions— row of interactive elementsinput— form fields in modalsdivider— visual separatorcontext— small metadata textheader— bold large text
Use https://app.slack.com/block-kit-builder for visual design.
5. Socket Mode vs HTTP
- Socket Mode (recommended for internal bots and local development): no public URL needed, uses a WebSocket
- Create an app-level token (
xapp-) with theconnections:writescope and enable Socket Mode in the app settings const app = new App({ token: process.env.SLACK_BOT_TOKEN, appToken: process.env.SLACK_APP_TOKEN, socketMode: true }), thenawait app.start()- Apps using Socket Mode cannot be published in the Slack Marketplace
- Create an app-level token (
- HTTP mode (for Marketplace/distributed apps): needs a public HTTPS URL for events, commands and interactivity (a tunnel such as ngrok works for dev)
const app = new App({ token: process.env.SLACK_BOT_TOKEN, signingSecret: process.env.SLACK_SIGNING_SECRET })await app.start(3000)
6. Deployment
- Simple: Railway, Render, or any Node.js host
- Production: Container on ECS/Cloud Run with health checks
- Serverless: AWS Lambda with Slack Bolt's
AwsLambdaReceiver
7. Rate Limits
Web API methods sit in tiers: Tier 1 is 1+ requests/min, Tier 2 20+/min, Tier 3 50+/min, Tier 4 100+/min (per method, per workspace and app). chat.postMessage is a special case: about 1 message per second per channel, with short bursts tolerated. Since 29 May 2025, new commercially distributed apps that are not Marketplace-approved get a much stricter limit on conversations.history and conversations.replies (1 request/min); check each method's page for its tier.
On HTTP 429, wait for the number of seconds in the Retry-After header before retrying. The Node WebClient queues and retries rate-limited calls on its own; in Python add RateLimitErrorRetryHandler to the client.
Examples
Example 1: Daily Standup Bot
Input: "Build a Slack bot that posts daily standup prompts at 9am, collects responses via thread, and summarizes them at 5pm."
Output: A Bolt.js app with:
- Cron job (node-cron) posting standup template to #team channel at 9:00
- Thread listener collecting responses, parsing "yesterday/today/blockers" format
- 5pm summary job that aggregates all thread replies into a formatted Block Kit message
/standup-skipslash command to mark days off
Example 2: Incident Alert Bot
Input: "Create a Slack bot that receives PagerDuty webhooks and creates an incident channel with pre-populated runbook links."
Output: A Bolt.js app with:
- HTTP endpoint receiving PagerDuty webhook payloads
- Auto-creates
#incident-{date}-{title}channel (needschannels:manageandpins:write) - Posts incident details with severity-colored sidebar
- Pins runbook links based on service name lookup
- Adds on-call responders to channel automatically
/incident-resolvecommand to archive channel and post timeline
Guidelines
- Always set
textfallback alongsideblocks(for notifications and accessibility) - Use ephemeral messages (
response_type: 'ephemeral') for user-specific responses - Store tokens in environment variables, never in code
- Implement proper error handling — Slack silently drops unacknowledged interactions after 3 seconds
- Always
await ack()within 3 seconds for commands and interactions - Use
app.error()global handler for uncaught errors - For long operations, ack immediately, then use
respond()orchat.postMessagelater - Test with Slack's Socket Mode locally before deploying
- Keep Block Kit messages under 50 blocks and 3000 characters per text field