Add Slackbot code

[smurching]

Add Slack Bot with OAuth Authentication

Overview

This PR introduces a Slack bot that provides personalized, authenticated access to Databricks AI agents. The bot implements a UX featuring ephemeral messages, sharing capabilities, and multi-user support.

Key Features

• Data privacy-respecting UX: Bot responds privately to users via ephemeral messages, with options to share responses publicly or dismiss them
• OAuth Authentication: Seamless Databricks OAuth flow with automatic response after authentication completes
• Multi-user Support: Each user authenticates independently and receives personalized responses
• Thread-based Conversations: All interactions happen in threads for clean channel organization
• Interactive Buttons:
  • "Post in thread" - Share private response publicly
  • "Dismiss" - Remove ephemeral message
  • "Generate my own answer" - Allow other users to get their own personalized responses
• Visual Feedback: Eyes emoji reaction on message receipt, white checkmark when shared
• Auto-respond After Auth: Bot stores pending messages and automatically responds after OAuth completes

Architecture

The bot uses Socket Mode for development and consists of:

• app_socket.py: Main Slack app with event handlers and button actions
• oauth_server.py: OAuth callback handler running on port 8000
• handlers/: Message processing and authentication logic
• state/: In-memory state management (can be replaced with Redis/DynamoDB for production)
• client/: Databricks API client for agent calls

User Flow

1. User mentions bot: @DatabricksBot hello
2. Bot reacts with 👀 emoji for immediate acknowledgment
3. First-time users: Bot sends ephemeral OAuth prompt in thread
4. After OAuth or for returning users: Bot sends "Working on it..." status message
5. Bot sends ephemeral response with "Post in thread" and "Dismiss" buttons (visible only to requesting user)
6. Bot posts public message with "Generate my own answer" button for other users
7. When shared: Ephemeral message disappears, response posted publicly, ✅ reaction added to original message

Technical Highlights

• Ephemeral Messages: Uses chat_postEphemeral for private responses
• Response URL Pattern: Uses respond(delete_original=True) for dismissing ephemeral messages
• Event Deduplication: Properly handles both app_mention and message events without duplication
• PKCE OAuth Flow: Secure OAuth implementation with proper state management
• Conversation History: Per-thread history for context-aware responses
• Regex Action Patterns: Dynamic action_id matching for button interactions

Testing

Run the bot locally:

cd slack-bot
python app_socket.py

The bot will:

• Connect to Slack via Socket Mode
• Start OAuth callback server on port 8000
• Log all events and interactions to console

Future Enhancements

• Support better parsing and rendering of agent output, especially of tool-calling agents (if a user query triggers an agent tool call e.g "What is the 1000th fibonacci number?", the Slack bot currently fails to parse the agent output)
• Support deploying as a Databricks app (add app.yaml, make the choice of serving endpoint parametrizable, add deployment instructions, etc)
• Replace in-memory state storage (of user oauth tokens) with something durable and secure (e.g. lakebase?) for production. Needs some discussion with security

Related Documentation

See slack-bot/README.md for:

• Complete setup instructions
• Slack app configuration
• Environment variables
• Development guide
• Troubleshooting tips