MS Teams RAG AI Bot using APIs

A production-ready Microsoft Teams bot that integrates with CustomGPT.ai API to bring your AI-powered knowledge base directly into Teams conversations.

Get you CustomGPT.ai RAG API key here, needed to use this integration.

More information on GitHub - https://github.com/Poll-The-People/customgpt-integrations

Quick Links: Prerequisites | Getting Credentials | Setup | Deployment | Confluence | Troubleshooting

Features
Prerequisites
Getting Credentials
Quick Setup
Configuration
Deployment
Confluence Integration
Usage
Troubleshooting
Support

Features

Core Capabilities

🤖 Intelligent Q&A: Access your CustomGPT knowledge base in Teams
💬 Context-Aware Conversations: Maintains conversation history
🧵 Threading Support: Seamless conversation threads
🎯 @Mentions: Responds to bot mentions in channels
🃏 Adaptive Cards: Rich, interactive UI elements
⚡ Slash Commands: /help, /reset, /status, /search

In personal chat: /help
In channel: @CustomGPT Bot 2 /help

Enterprise Features

🔐 Azure AD Authentication: Secure enterprise authentication
📋 Audit Logging: Track all bot interactions
🛡️ Access Control: Tenant, channel, and user restrictions
🔄 Rate Limiting: Multi-tier (user/channel/tenant) protection
📊 Analytics: Application Insights integration
🌍 Multi-tenant: Works across organizations

Optional Integrations

🔍 Confluence: Search Atlassian Confluence directly from Teams
📎 File Attachments: Process documents and images
💼 Meeting Support: Works in Teams meetings

Prerequisites

Requirement	Free Tier Available	Purpose
Azure Account	✅ Yes	Bot registration and hosting
CustomGPT.ai Account	✅ Trial	AI knowledge base API
Microsoft Teams Admin	-	Install bot in Teams
Python 3.8 - 3.11	✅ Yes	Local development
Redis (Optional)	✅ Yes	Distributed rate limiting

Getting Credentials

You need 4 required credentials to run the bot. Follow these steps to get them.

1. Microsoft Teams Credentials (Azure Portal)

Step 1: Create App Registration

Go to Azure Portal
Search for "Microsoft Entra ID" → "App registrations" → "+ New registration"

Fill in:

Name: customgpt-teams-bot
Supported account types: Single tenant (recommended) or Multi-tenant
Redirect URI: Leave blank

Click "Register"
COPY THE APPLICATION (CLIENT) ID - Save this as your TEAMS_APP_ID

Step 2: Create Client Secret

In the left menu, click "Certificates & secrets" → "Client secrets" tab
Click "+ New client secret"
Description: Teams Bot Secret, Expires: 24 months
Click "Add"
IMMEDIATELY COPY THE VALUE - Save this as your TEAMS_APP_PASSWORD

⚠️ Copy the Value, NOT the "Secret ID". You can only see it once!

Step 3: Create Azure Bot Resource

Azure Portal → "Create a resource" → "Azure Bot"
Fill in:
- Bot handle: customgpt-teams-bot
- Pricing tier: F0 (Free)
- Microsoft App ID: Enter the ID from Step 1
"Review + Create" → "Create"
You'll add the messaging endpoint later when you deploy

Step 4: Enable Teams Channel

Go to your Azure Bot → "Channels" → "Microsoft Teams" icon → "Apply"

✅ Credentials ready:

TEAMS_APP_ID from Step 1
TEAMS_APP_PASSWORD from Step 2

2. CustomGPT Credentials

Get CUSTOMGPT_API_KEY

Go to CustomGPT Dashboard
Sign in to your account
Click profile icon (top right) → "API Settings"
Click "Generate API Key" if you don't have one
Name it: Teams Bot
Copy the API Key

Get CUSTOMGPT_PROJECT_ID

Get this from https://app.customgpt.ai/ > Select your agent > Deploy > "See top right section"

3. Optional: Confluence Credentials

Only needed if you want Confluence search integration.

For Confluence Cloud

Base URL: Your Confluence URL

Example: https://your-company.atlassian.net

Create API Token:
- Go to Atlassian API Tokens
- Click "Create API token"
- Label: Teams Bot
- Copy the token
Username: Your Atlassian email address

For Confluence Server/Data Center

Base URL: Your self-hosted URL

Example: https://confluence.yourcompany.com

Create Personal Access Token:
- Profile → Personal Access Tokens
- Create token: Teams Bot
- Copy the token
Username: Your Confluence username (not email)

Quick Setup

Step 1: Clone and Install

# Clone repository
git clone <your-repo>
cd "customgpt-integrations/MS Teams"

# Create virtual environment
python3 -m venv venv

# Activate virtual environment
source venv/bin/activate  # macOS/Linux
# OR
venv\Scripts\activate     # Windows

# Install dependencies
pip install --upgrade pip
pip install -r requirements.txt

Step 2: Configure Environment

# Copy environment template
cp .env.example .env

# Edit .env with your credentials
nano .env  # or use any text editor

Required Configuration:

# Microsoft Teams (from Azure Portal)
TEAMS_APP_ID=12345678-1234-1234-1234-123456789abc
TEAMS_APP_PASSWORD=abc123~XYZ456.789def-GHI012_JKL345
TEAMS_APP_TYPE=MultiTenant

# CustomGPT (from CustomGPT Dashboard)
CUSTOMGPT_API_KEY=xxxxxxxxxxxxxxxxxxxxxxxxxxxxx
CUSTOMGPT_PROJECT_ID=12345

Optional - Confluence Integration:

# Confluence (only if integrating)
CONFLUENCE_ENABLED=false  # Set to true to enable
CONFLUENCE_BASE_URL=https://your-company.atlassian.net
CONFLUENCE_IS_CLOUD=true
[email protected]
CONFLUENCE_API_TOKEN=your-api-token

Step 3: Start the Bot

# Make sure virtual environment is activated
source venv/bin/activate

# Start the bot
python3 app.py

Expected Output:

INFO - CustomGPT client initialized
INFO - Rate limiter initialized
INFO - Conversation manager initialized
INFO - Bot initialization complete
 * Running on http://0.0.0.0:3978

Step 4: Verify Health

Open another terminal:

curl http://localhost:3978/health

Expected Response:

{
  "status": "healthy",
  "bot": "CustomGPT Teams Bot",
  "version": "1.0.0"
}

✅ Your bot is now running locally!

Configuration

Rate Limiting

Control request rates to prevent abuse:

RATE_LIMIT_PER_USER=20      # Messages per minute per user
RATE_LIMIT_PER_CHANNEL=100  # Messages per hour per channel
RATE_LIMIT_PER_TENANT=500   # Messages per hour per organization

Security & Access Control

Restrict bot access by tenant, channel, or user:

# Restrict to specific organizations (comma-separated)
ALLOWED_TENANTS=tenant-id-1,tenant-id-2

# Restrict to specific channels
ALLOWED_CHANNELS=channel-id-1,channel-id-2

# Block specific users
BLOCKED_USERS=user-id-1,user-id-2

Conversation Settings

Control conversation behavior:

CONVERSATION_TIMEOUT=86400   # 24 hours (in seconds)
MAX_CONTEXT_MESSAGES=10      # Number of messages to keep in context
ENABLE_THREADING=true        # Enable thread support

Bot Behavior

Customize bot responses:

MAX_MESSAGE_LENGTH=4000           # Max message length
SHOW_CITATIONS=true               # Show source citations
REQUIRE_MENTION_IN_CHANNELS=true  # Require @mention in channels
ENABLE_ADAPTIVE_CARDS=true        # Use rich cards
DEFAULT_LANGUAGE=en               # Default response language
RESPONSE_TIMEOUT=30               # API timeout in seconds

Logging

LOG_LEVEL=INFO  # DEBUG for development, INFO for production, WARNING for minimal logs

Server Settings

PORT=3978      # Default Teams bot port
HOST=0.0.0.0   # Listen on all interfaces

Deployment

Local Development with Ngrok

For testing with Teams while developing locally:

Install ngrok:

# macOS
brew install ngrok

# Or download from https://ngrok.com

Start your bot:
```
source venv/bin/activate
python3 app.py
```
In another terminal, start ngrok:
```
ngrok http 3978
```
Copy the HTTPS URL from ngrok output:
- Look for the "Forwarding" line in ngrok output
- Copy the HTTPS URL (e.g., https://631af35e4e38.ngrok-free.app)
- Important: Add /api/messages to the end
Update Azure Bot messaging endpoint:
- Go to Azure Portal → Your Bot → Configuration
- Find the Messaging endpoint field
- Paste: https://your-ngrok-url.ngrok-free.app/api/messages
- Example: https://631af35e4e38.ngrok-free.app/api/messages
- Click "Apply"
Test in Teams - Your bot is now accessible!

⚠️ Important Notes:

The /api/messages path is required - this is where your bot receives messages
Free ngrok URLs change every time you restart ngrok
You'll need to update the Azure Bot messaging endpoint each time ngrok restarts
For persistent URLs, consider ngrok paid plan or deploy to production

Production Deployment Options

Option 1: Azure Web App (Recommended)

Best for: Production deployments with enterprise features

Steps:

Create Azure Web App:

az webapp up \
  --name customgpt-teams-bot \
  --resource-group customgpt-rg \
  --runtime "PYTHON:3.11" \
  --sku B1

Configure Environment Variables:
- Azure Portal → Web App → Configuration → Application settings
- Add all variables from your .env file
- Click "Save"
Update Bot Endpoint:
- Azure Portal → Bot → Configuration
- Messaging endpoint: https://customgpt-teams-bot.azurewebsites.net/api/messages
Enable Application Insights (optional but recommended):
- Add APPLICATION_INSIGHTS_KEY to app settings

Cost: ~$13/month for B1 tier (or ~$55/month for S1)

Option 2: Docker Container

Best for: Self-hosted or cloud-agnostic deployments

Build and Run:

# Build image
docker build -t customgpt-teams-bot .

# Run container
docker run -d \
  --name teams-bot \
  -p 3978:3978 \
  --env-file .env \
  customgpt-teams-bot

# View logs
docker logs -f teams-bot

Docker Compose:

# Run with Docker Compose
docker-compose up -d

# With Redis for rate limiting
docker-compose --profile with-redis up -d

Option 3: Free Cloud Hosting

Railway.app (Recommended for beginners):

Fork this repository
Go to Railway.app
Create new project from GitHub
Add environment variables
Deploy automatically
Copy public URL → Update Azure bot endpoint

Free Tier: 500 hours/month, $5 credit

Render.com:

Create Web Service
Connect GitHub repo
Set environment variables
Deploy

Free Tier: 750 hours/month

Install in Microsoft Teams

Step 1: Create Teams App Package

Update manifest:
- Edit deployment/manifest.json
- Find the botId field (around line 28)
- Replace YOUR-BOT-ID-HERE with your TEAMS_APP_ID (the Microsoft App ID from your .env file)
- Also update webApplicationInfo.id with the same ID
- Update name, description, etc. as needed
Example:
```
"bots": [
    {
        "botId": "12345678-1234-1234-1234-123456789abc",
```
What is botId?
- It's your Microsoft App ID (same as TEAMS_APP_ID in .env)
- It uniquely identifies your bot to Microsoft Teams
- Teams uses it to route messages to your bot
- It must match the App ID registered in Azure
Create icons:
- color.png: 192x192px color icon
- outline.png: 32x32px outline icon
- Place in deployment/ folder

Create ZIP package:

cd deployment
zip -r customgpt-bot.zip manifest.json color.png outline.png

Step 2: Upload to Teams

Method 1: Direct Upload (Simple)

Open Microsoft Teams
Go to Apps → Manage your apps → Upload an app
Choose "Upload a custom app"
Select customgpt-bot.zip
Click "Add"

Method 2: Teams Developer Portal (Recommended for troubleshooting duplicates)

If you encounter duplicate app errors or upload issues:

Open Teams Developer Portal:
- Visit: https://dev.teams.microsoft.com/apps
- Or in Teams: Apps → Developer Portal
Import your app:
- Click "Import app"
- Upload your customgpt-bot.zip
- The portal will detect duplicates and let you overwrite or create new
Manage app details:
- Edit app information in the portal if needed
- Preview how it will appear in Teams
Publish:
- Click "Publish" → "Publish to org"
- App will be available in your Teams organization

Benefits of Developer Portal:

Better handling of duplicate apps
Visual editor for manifest properties
App validation and testing tools
Version management

Step 3: Test the Bot

Personal Chat: Send a direct message to the bot
Channel: @mention the bot: @CustomGPT Bot hello
Commands: Try /help to see available commands

✅ Your bot is now live in Teams!

Confluence Integration

Enable Confluence Search

Add Confluence search capability to your Teams bot, allowing users to search and access Confluence pages directly from Teams.

Setup

Get Confluence credentials (see Getting Credentials)

Update .env file:

CONFLUENCE_ENABLED=true
CONFLUENCE_BASE_URL=https://your-company.atlassian.net
CONFLUENCE_IS_CLOUD=true
[email protected]
CONFLUENCE_API_TOKEN=your-confluence-api-token
CONFLUENCE_DEFAULT_SPACE=  # Optional: default space key
CONFLUENCE_SEARCH_LIMIT=5  # Max results to return

Restart the bot:
```
python3 app.py
```
Verify in logs:
```
INFO - Confluence client initialized
```

Usage

Commands:

/search <query> - Search Confluence pages

/search API documentation
/search onboarding process

/confluence - Show integration status and help

Confluence Integration Active

Available Spaces:
• Engineering (ENG)
• Documentation (DOCS)

Commands:
• /search <query> - Search Confluence
• /confluence - Show this help

Features:

Rich Adaptive Cards with search results
Direct links to Confluence pages
Content excerpts and metadata
Respects Confluence permissions

Authentication Methods

For Confluence Cloud (Recommended):

[email protected]
CONFLUENCE_API_TOKEN=YOUR_API_TOKEN

For Confluence Server/Data Center:

CONFLUENCE_IS_CLOUD=false
CONFLUENCE_USERNAME=your-username  # Not email
CONFLUENCE_API_TOKEN=YOUR_PERSONAL_ACCESS_TOKEN

Troubleshooting Confluence

Authentication Errors:

# Test credentials manually
curl -u "[email protected]:YOUR_TOKEN" \
  https://your-domain.atlassian.net/wiki/rest/api/space

No Results:

Check space permissions
Verify bot user has read access
Try broader search terms

Timeout Errors:

Increase RESPONSE_TIMEOUT in .env
Reduce CONFLUENCE_SEARCH_LIMIT

Usage

Basic Commands

Command	Description	Example
`/help`	Show available commands	`/help`
`/start`	Start new conversation	`/start`
`/reset`	Clear conversation context	`/reset`
`/status`	Check rate limits and bot status	`/status`
`/search`	Search Confluence (if enabled)	`/search API docs`
`/confluence`	Show Confluence integration status	`/confluence`

Interacting with the Bot

Personal Chat:

You: What is our refund policy?
Bot: According to our documentation...

In Channels (requires @mention):

You: @CustomGPT Bot what are the office hours?
Bot: Our office hours are...

In Threads:

You: Tell me about product features
Bot: Here are our key features...
You: (reply in thread) What about pricing?
Bot: (continues in thread) Our pricing is...

Adaptive Cards

The bot uses rich Adaptive Cards for:

Welcome messages with quick actions
Response cards with citations
Search results with direct links
Error messages with helpful tips

Rate Limits

Check your current rate limit status:

/status

Rate Limits:
• User: 15/20 requests this minute
• Channel: 45/100 requests this hour
• Tenant: 230/500 requests this hour

Bot Status: ✅ Healthy

Troubleshooting

Common Issues

1. "Configuration error: Missing required configuration"

Cause: Missing or empty credentials in .env file

Solution:

Check .env file exists in project directory
Verify all required fields are filled:
- TEAMS_APP_ID
- TEAMS_APP_PASSWORD
- CUSTOMGPT_API_KEY
- CUSTOMGPT_PROJECT_ID
No extra spaces or quotes around values
Restart bot after editing .env

2. Bot Not Responding in Teams

Possible Causes & Solutions:

✅ Check messaging endpoint:

Azure Portal → Bot → Configuration
Verify endpoint is correct: https://your-domain.com/api/messages
For local dev with ngrok, update with ngrok HTTPS URL

✅ Verify bot is installed:

Teams → Apps → Manage your apps
Confirm bot is installed

✅ Check bot is added to channel:

For channel conversations, bot must be added as member
Try personal chat first to isolate issue

✅ View bot logs:

# Local
# Check terminal output

# Docker
docker logs customgpt-teams-bot

# Azure
az webapp log tail --name your-bot --resource-group your-rg

3. Authentication Errors

Symptom: Authentication failed in logs

Solution:

Verify App ID matches Azure Portal:

# Check .env file
cat .env | grep TEAMS_APP_ID

# Compare with Azure Portal → Bot → Configuration

Regenerate client secret if needed:
- Azure Portal → Bot → Configuration → Manage
- Certificates & secrets → New client secret
- Update TEAMS_APP_PASSWORD in .env
Restart bot after updating credentials

4. CustomGPT API Errors

Symptom: Bot responds with "Error processing request"

Test API manually:

curl -X POST https://app.customgpt.ai/api/v1/projects/YOUR_PROJECT_ID/conversations \
  -H "Authorization: Bearer YOUR_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"prompt": "test"}'

If test fails:

Regenerate API key in CustomGPT dashboard
Verify project ID is correct
Check project is active (not paused)
Update .env and restart bot

5. Rate Limit Issues

Symptom: "Rate limit exceeded" messages

Solutions:

Increase limits in .env:

RATE_LIMIT_PER_USER=30      # Increase from 20
RATE_LIMIT_PER_CHANNEL=200  # Increase from 100

Use Redis for distributed rate limiting:
```
REDIS_URL=redis://localhost:6379
```
Check current status:
```
/status
```

6. Python Import Errors

Symptom: ModuleNotFoundError when starting bot

Solution:

# Ensure virtual environment is activated
source venv/bin/activate

# Verify activation (should see "(venv)" in prompt)

# Reinstall dependencies
pip install -r requirements.txt

# Verify bot module imports
python3 -c "import bot; print('✅ Success')"

Debug Mode

Enable detailed logging:

LOG_LEVEL=DEBUG

Restart bot and check logs for detailed information about:

HTTP requests/responses
Authentication attempts
API calls
Error stack traces

Health Check

Verify bot is running:

curl http://localhost:3978/health

Healthy Response:

{
  "status": "healthy",
  "bot": "CustomGPT Teams Bot",
  "version": "1.0.0",
  "timestamp": "2025-11-04T00:00:00.000Z"
}

Unhealthy Response:

No response: Bot not running or port blocked
Error response: Check logs for details

Getting Help

Before asking for help, collect:

Bot logs (last 50 lines with errors)
Your .env configuration (REDACT secrets!)
Python version: python3 --version
Steps to reproduce the issue

Development

Running Tests

# Activate virtual environment
source venv/bin/activate

# Run all tests
pytest

# Run with coverage
pytest --cov=. --cov-report=html

# Run specific test
pytest tests/test_bot.py -v

Code Quality

# Format code
black *.py

# Lint code
flake8 *.py

# Type check
mypy *.py

Security Best Practices

1. Credential Management

✅ Never commit secrets:

# .gitignore already includes:
.env
*.token
credentials.json

✅ Use Azure Key Vault for production:

az keyvault secret set \
  --vault-name your-vault \
  --name ConfluenceApiToken \
  --value "YOUR_TOKEN"

✅ Rotate secrets regularly (every 3-6 months)

2. Access Control

✅ Enable tenant restrictions in production:

ALLOWED_TENANTS=your-tenant-id

✅ Use service accounts for Confluence (not personal accounts)

✅ Audit logging enabled:

ENABLE_AUDIT_LOGGING=true

3. Network Security

✅ Always use HTTPS ✅ Configure firewalls to allow only necessary traffic ✅ Use Azure Network Security Groups for production

4. Data Privacy

✅ No message storage by default ✅ Conversation timeout clears context:

CONVERSATION_TIMEOUT=86400  # 24 hours

Support

CustomGPT Resources

Landing Page: https://customgpt.ai
Dashboard: https://app.customgpt.ai
API Docs: https://docs.customgpt.ai/api-reference
Postman Collection: https://www.postman.com/customgpt/customgpt/overview
Office Hours: https://calendly.com/pollthepeople/office-hours
YouTube: https://www.youtube.com/channel/UC6HOk7Z9OwVPNYiC7SKMJ6g

Microsoft Resources

Azure Bot Service: https://docs.microsoft.com/azure/bot-service
Teams Platform: https://docs.microsoft.com/microsoftteams/platform
Bot Framework: https://github.com/microsoft/botbuilder-python

Community

GitHub Issues: Report bugs or request features
Discussions: Ask questions and share tips

License

See main repository LICENSE file.

Table of Contents

Features

Core Capabilities

Enterprise Features

Optional Integrations

Prerequisites

Getting Credentials

1. Microsoft Teams Credentials (Azure Portal)

Step 1: Create App Registration

Step 2: Create Client Secret

Step 3: Create Azure Bot Resource

Step 4: Enable Teams Channel

2. CustomGPT Credentials

Get CUSTOMGPT_API_KEY

Get CUSTOMGPT_PROJECT_ID

3. Optional: Confluence Credentials

For Confluence Cloud

For Confluence Server/Data Center

Quick Setup

Step 1: Clone and Install

Step 2: Configure Environment

Step 3: Start the Bot

Step 4: Verify Health

Configuration

Rate Limiting

Security & Access Control

Conversation Settings

Bot Behavior

Logging

Server Settings

Deployment

Local Development with Ngrok

Production Deployment Options

Option 1: Azure Web App (Recommended)

Option 2: Docker Container

Option 3: Free Cloud Hosting

Install in Microsoft Teams

Step 1: Create Teams App Package

Step 2: Upload to Teams

Step 3: Test the Bot

Confluence Integration

Enable Confluence Search

Setup

Usage

Authentication Methods

Troubleshooting Confluence

Usage

Basic Commands

Interacting with the Bot

Adaptive Cards

Rate Limits

Troubleshooting

Common Issues

1. "Configuration error: Missing required configuration"

2. Bot Not Responding in Teams

3. Authentication Errors

4. CustomGPT API Errors

5. Rate Limit Issues

6. Python Import Errors

Debug Mode

Health Check

Getting Help

Development

Running Tests

Code Quality

Security Best Practices

1. Credential Management

2. Access Control

3. Network Security

4. Data Privacy

Support

CustomGPT Resources

Microsoft Resources

Community

License