PMBot - Product Manager Assistant 🤖

A sophisticated AI-powered chatbot designed to assist Product Managers with RFE (Request for Enhancement) creation, validation, and JIRA integration. Built for Red Hat Enterprise context but adaptable to other organizations. Features Streamlit frontend, FastAPI backend, and deployed on OpenShift.

🌐 Live Demo

Try it now - no setup required! (Live access available until Friday, August 15th, 2025)

🎯 Streamlit Web App: https://pm-chatbot.apps.cluster-znvdr.znvdr.sandbox203.opentlc.com
🔧 API Endpoint: https://pm-chatbot-api.apps.cluster-znvdr.znvdr.sandbox203.opentlc.com
📚 API Documentation: https://pm-chatbot-api.apps.cluster-znvdr.znvdr.sandbox203.opentlc.com/docs

✨ Key Features

🎯 RFE Generation: Create well-structured Request for Enhancement documents
📋 Template Validation: Ensure RFEs follow proper guidelines and formats
🔗 JIRA Integration: Seamlessly create and manage JIRA tickets
📚 Document Search: Query Red Hat AI documentation with vector-based search
🌐 REST API: Programmable access to all functionality
🤖 MCP Support: Model Context Protocol integration for AI agents and tools
☁️ Cloud Ready: Containerized and deployed on OpenShift

🚀 Getting Started

Option 1: Use the Web Interface (Recommended)

Visit the Streamlit Web App
Start chatting with the PM assistant
Ask for help with RFE creation, JIRA integration, or Red Hat AI documentation

Option 2: Use the REST API

Check out the API Documentation
Test endpoints directly in the browser

Option 3: MCP Integration (For AI Agents & Tools)

Model Context Protocol (MCP) allows AI agents and tools to interact with PMBot programmatically. Perfect for integrating with Claude Desktop, VS Code extensions, or custom AI workflows.

Quick MCP Setup

Add this to your MCP client configuration (e.g., Claude Desktop's mcp.json):

{
  "pm-chatbot-production": {
    "command": "mcp-proxy",
    "args": [
      "https://pm-chatbot-api.apps.cluster-znvdr.znvdr.sandbox203.opentlc.com/mcp"
    ],
    "env": {
      "AUTHORIZATION": "Bearer pmbot-production-token"
    }
  }
}

Available MCP Tools

mcp_rfe_generate (POST /mcp/rfe/generate) - Generate RFEs from natural language requests
mcp_rfe_validate (POST /mcp/rfe/validate) - Validate RFE content against guidelines
mcp_search_documents (POST /mcp/documents/search) - Search Red Hat AI documentation
mcp_get_models (GET /mcp/models) - List available AI models
mcp_create_jira_issue (POST /mcp/jira/issues) - Create JIRA tickets
mcp_update_jira_issue (PUT /mcp/jira/issues/{issue_key}) - Update existing JIRA tickets

Note: For direct API access, use the /api/v1/ endpoints shown in the API examples below.

MCP Usage Examples

Once configured, AI agents can use natural language like:

"Generate an RFE for improving API rate limiting"
"Validate this RFE draft against Red Hat guidelines"
"Search for Red Hat AI installation requirements"
"What models are available for RFE generation?"
"Create a JIRA ticket for this enhancement request"
"Update JIRA issue RHOAI-123 with additional details"

MCP Endpoint: https://pm-chatbot-api.apps.cluster-znvdr.znvdr.sandbox203.opentlc.com/mcp

🛠️ API Usage Examples

Generate an RFE

# With authentication token (required for deployed API)
curl -X POST "https://pm-chatbot-api.apps.cluster-znvdr.znvdr.sandbox203.opentlc.com/api/v1/rfe/generate" \
  -H "Content-Type: application/json" \
  -H "Authorization: Bearer pmbot-production-token" \
  --max-time 120 \
  -d '{
    "prompt": "Create an RFE for improving AI model performance monitoring",
    "context": "Red Hat AI platform needs better monitoring capabilities for deployed models",
    "selected_product": "Red Hat AI"
  }'

Search Documentation

curl -X POST "https://pm-chatbot-api.apps.cluster-znvdr.znvdr.sandbox203.opentlc.com/mcp/documents/search" \
  -H "Content-Type: application/json" \
  -H "Authorization: Bearer pmbot-production-token" \
  -d '{
    "query": "Red Hat AI installation requirements",
    "limit": 5
  }'

🏗️ Architecture

            ┌─────────────────┐
            │   OpenShift     │
        ┌── │   Deployment    │ ──┐
        │   │                 │   │
        ▼   └─────────────────┘   ▼
┌─────────────────┐    ┌─────────────────┐    
│   Streamlit     │    │    FastAPI      │    
│   Frontend      │◄──►│    Backend      │
│                 │    │  + MCP Server   │    
└─────────────────┘    └─────────────────┘\    
         │                       │         \            
         ▼                       ▼          \           
┌─────────────────┐    ┌─────────────────┐   \┌─────────────────┐
│   User Chat     │    │  Vector DB      │    │   JIRA API      │
│   Interface     │◄──►│  (FAISS)        │◄──►│   Integration   │
└─────────────────┘    └─────────────────┘    └─────────────────┘
         │                       │                       
         ▼                       ▼                       
┌─────────────────┐    ┌─────────────────┐               
│  MCP Clients    │    │   AI Agents     │               
│ (Claude, Tools) │    │  & Extensions   │               
└─────────────────┘    └─────────────────┘

📁 Project Structure

PMBot/
├── 🎯 Core Components
│   ├── pm_chatbot_main.py      # Streamlit frontend application
│   ├── api_server.py           # FastAPI backend server
│   ├── atlassian_client.py     # JIRA/Confluence integration
│   ├── rfe_manager.py          # RFE creation and validation
│   ├── document_processor.py   # Document processing and indexing
│   ├── vector_database.py      # Vector database operations
│   └── auth.py                 # Authentication utilities
│
├── 📚 Documentation & Data
│   ├── documents/              # Source documentation (PDFs)
│   ├── document_cache/         # Processed document cache (JSON)
│   ├── vector_db/              # FAISS vector database files
│   └── docs/                   # Project documentation
│       └── RFE_Issue_Description_Guidelines.md
│
├── ☁️ Deployment & Infrastructure
│   ├── deployment/
│   │   ├── docker/             # Container definitions
│   │   ├── openshift/          # OpenShift deployment configs
│   │   └── scripts/            # Deployment scripts
│   ├── .github/                # GitHub Actions workflows
│   └── deploy-to-cluster.sh    # Main deployment script
│
├── 🔧 Configuration
│   ├── config/
│   │   ├── requirements.txt        # Core Python dependencies
│   │   ├── requirements.api.txt    # API-specific dependencies
│   │   ├── requirements.full.txt   # Complete dependency list
│   │   └── nginx.conf              # Nginx configuration
│   ├── .env.template               # Environment variables template
│   └── .streamlit/                 # Streamlit configuration
│
└── 🛠️ Utilities & Scripts
    ├── scripts/
    │   └── setup-github-secrets.sh # GitHub secrets setup
    └── utils/
        ├── local-app.sh            # Local Streamlit app launcher
        └── local-backend.sh        # Local API server launcher

🔧 Local Development Setup

Prerequisites

Python 3.11+ installed
Git installed
JIRA account with personal access token
MaaS (Model-as-a-Service) API credentials

System Dependencies

macOS:

# Install SWIG (required for FAISS vector database)
brew install swig

Ubuntu/Debian:

# Install SWIG and build tools
sudo apt-get update
sudo apt-get install swig build-essential

RHEL/CentOS/Fedora:

# Install SWIG and development tools
sudo yum install swig gcc-c++ python3-devel
# OR for newer versions:
sudo dnf install swig gcc-c++ python3-devel

🚀 Quick Start (5 minutes - Recommended!)

Step 1: Clone & Setup

# Clone the repository
git clone https://github.com/jhuang2026/PMBot.git
cd PMBot

Step 2: Configure Environment

# Copy the template
cp .env.template .env

# Edit with your credentials
nano .env  # or use your preferred editor

Step 3: Get Your Credentials

JIRA Personal Access Token

Go to https://issues.redhat.com/secure/ViewProfile.jspa?selectedTab=com.atlassian.pats.pats-plugin:jira-user-personal-access-tokens
Create a new token
Copy it to JIRA_PERSONAL_TOKEN in your .env file

MaaS API Credentials For Red Hat users, visit the Red Hat MaaS Platform to get your credentials.

Contact your AI service provider for:

API Key
Base URL endpoint
Model name

Add at least ONE model configuration to your .env file.

Step 4: Run the Application

# Start the application (installs dependencies automatically)
./utils/local-app.sh

# Note: If you get FAISS installation errors, install SWIG first:
# macOS: brew install swig
# Ubuntu: sudo apt-get install swig build-essential
# RHEL/CentOS: sudo yum install swig gcc-c++ python3-devel

The app will be available at: http://localhost:8501

Step 5: (Optional) Start API Server

# In a new terminal, start the API server
./utils/local-backend.sh

# Note: This installs additional API dependencies from requirements.api.txt

API will be available at: http://localhost:8000
API docs at: http://localhost:8000/docs

Environment Variables Quick Reference

Required Variables

# Minimum required for basic functionality
MAAS_PHI_4_API_KEY=your_api_key
MAAS_PHI_4_BASE_URL=https://your-endpoint.com/v1
MAAS_PHI_4_MODEL_NAME=microsoft/phi-4

JIRA_URL=https://issues.redhat.com/
JIRA_PERSONAL_TOKEN=your_token

🚀 Production Deployment

Manual Deployment (Foundation)

Prerequisites for Deployment

OpenShift/Kubernetes cluster with admin access
Container registry access (Quay.io recommended)
Environment configuration (see below)

Container Registry Setup

⚠️ IMPORTANT REGISTRY NOTICE: This repository is currently configured to use quay.io/rh-ee-jashuang as the default registry. You should set up your own registry to avoid pushing images to someone else's account.

Option 1: Quay.io (Recommended)

# 1. Create account at https://quay.io
# 2. Create repository: quay.io/your-username/pm-chatbot-streamlit
# 3. Create repository: quay.io/your-username/pm-chatbot-api
# 4. Make repositories public for easier access
# 5. Login locally
podman login quay.io

Option 2: Docker Hub

# 1. Create account at https://hub.docker.com
# 2. Create repositories for streamlit and api images
# 3. Login locally
podman login docker.io

Using the Deploy Script

The deploy-to-cluster.sh script is a comprehensive, intelligent deployment tool that handles everything automatically. Perfect for first-time users!

⚠️ IMPORTANT: By default, this script is configured to use the quay.io/rh-ee-jashuang registry. You should change this to your own registry before deploying to avoid pushing to someone else's container registry.

Basic Usage (Simplest):

# Clone the repository
git clone https://github.com/jhuang2026/PMBot.git
cd PMBot

# Configure your environment
cp .env.template .env
# Edit .env with your credentials (see Step 3 above for credential sources)

# IMPORTANT: Use your own registry (recommended)
./deploy-to-cluster.sh -r quay.io/your-username

# OR: Deploy with default registry (uses rh-ee-jashuang - not recommended)
./deploy-to-cluster.sh

Advanced Usage Options:

# Deploy to a different cluster (specify API URL)
./deploy-to-cluster.sh -a https://api.your-cluster.com:6443

# Use your own container registry
./deploy-to-cluster.sh -r quay.io/your-username

# Deploy to a custom namespace
./deploy-to-cluster.sh -n my-custom-namespace

# Combine multiple options
./deploy-to-cluster.sh -a https://api.new-cluster.com:6443 -r quay.io/myorg -n production

# See all available options
./deploy-to-cluster.sh --help

🎯 What the Script Does Automatically:

🔍 Cluster Detection - Works with both new and existing OpenShift clusters
🏗️ Smart Building - Only rebuilds images when source code changes (saves time!)
📦 Container Management - Handles Podman setup, registry login, and image pushing
🔐 Security Setup - Creates all required secrets from your .env file
🚀 Deployment - Applies all OpenShift configurations and waits for readiness
📊 Cache Upload - Uploads pre-built RAG cache for faster startup
✅ Verification - Checks all services and provides final URLs

📋 First-Time User Checklist:

Before running the script, ensure you have:

✅ OpenShift cluster access (login with oc login)
✅ Container registry account (Quay.io or Docker Hub)
✅ Podman installed (script will try to start it automatically)
✅ Environment configured (.env file with your credentials)

⚡ Performance Features:

The script includes several optimizations for faster deployments:

Smart caching - Remembers previous deployments and skips unchanged components
Parallel operations - Builds and deploys multiple components simultaneously
Layer reuse - Docker layers are cached between builds
Incremental updates - Only updates what actually changed

🔧 Troubleshooting:

If deployment fails:

# Check your cluster connection
oc whoami
oc get nodes

# Verify Podman is working  
podman info

# Check container registry login
podman login quay.io  # or your registry

# See detailed script help
./deploy-to-cluster.sh --help

🌐 Deployment Scenarios:

# Scenario 1: RECOMMENDED - Deploy with your own registry
./deploy-to-cluster.sh -r quay.io/your-username

# Scenario 2: Deploy to a different OpenShift cluster with your registry
./deploy-to-cluster.sh -a https://api.cluster-abc.sandbox.com:6443 -r quay.io/your-username

# Scenario 3: Use Docker Hub instead of Quay
./deploy-to-cluster.sh -r docker.io/your-username

# Scenario 4: Deploy to a custom namespace with your registry
./deploy-to-cluster.sh -r quay.io/your-username -n my-custom-namespace

# Scenario 5: Production deployment (custom cluster + your registry + namespace)
./deploy-to-cluster.sh \
  -a https://api.prod-cluster.company.com:6443 \
  -r quay.io/your-company \
  -n pmbot-production

# Scenario 6: Development environment
./deploy-to-cluster.sh \
  -a https://api.dev-cluster.sandbox.com:6443 \
  -r quay.io/dev-team \
  -n pmbot-dev

# ⚠️ NOT RECOMMENDED: Deploy with default registry (uses rh-ee-jashuang)
# ./deploy-to-cluster.sh

💡 Pro Tips:

The script remembers your last deployment settings for faster re-deployments
Use -a flag to easily switch between different OpenShift clusters
Each deployment creates cache files to speed up subsequent deployments
The script works with any OpenShift-compatible cluster (OKD, RHOKS, etc.)

GitHub Actions Deployment (Automated)

This repository includes automated GitHub Actions workflows for building and deploying to OpenShift. This builds on the manual deployment foundation above. IMPORTANT: You must set up GitHub Secrets before deploying.

🔐 Security Setup for GitHub Deployment

⚠️ CRITICAL: Setting Up GitHub Secrets

This repository requires sensitive credentials (API keys, tokens) that must NOT be stored in the code. Instead, they should be configured as GitHub Secrets for secure deployment.

Step 1: Navigate to Repository Secrets

Go to your GitHub repository
Click Settings (in the repository menu)
In the left sidebar, click Secrets and variables → Actions
Click New repository secret

Step 2: Add Required Secrets

You need to add the following secrets (use the values from your .env file):

JIRA Configuration

JIRA_URL - Your JIRA instance URL (e.g., https://issues.redhat.com/)
JIRA_PERSONAL_TOKEN - Your personal JIRA API token

MaaS Model Configuration

For each model you want to use, add these secrets:

DeepSeek R1 Qwen 14B:

MAAS_DEEPSEEK_R1_QWEN_14B_API_KEY
MAAS_DEEPSEEK_R1_QWEN_14B_BASE_URL
MAAS_DEEPSEEK_R1_QWEN_14B_MODEL_NAME

Phi-4:

MAAS_PHI_4_API_KEY
MAAS_PHI_4_BASE_URL
MAAS_PHI_4_MODEL_NAME

Granite 3.3 8B Instruct:

MAAS_GRANITE_3_3_8B_INSTRUCT_API_KEY
MAAS_GRANITE_3_3_8B_INSTRUCT_BASE_URL
MAAS_GRANITE_3_3_8B_INSTRUCT_MODEL_NAME

Llama 4 Scout 17B:

MAAS_LLAMA_4_SCOUT_17B_API_KEY
MAAS_LLAMA_4_SCOUT_17B_BASE_URL
MAAS_LLAMA_4_SCOUT_17B_MODEL_NAME

Mistral Small 24B:

MAAS_MISTRAL_SMALL_24B_API_KEY
MAAS_MISTRAL_SMALL_24B_BASE_URL
MAAS_MISTRAL_SMALL_24B_MODEL_NAME

Step 3: Automated Setup (Recommended)

Use the provided script to set up all secrets easily:

# Install GitHub CLI if not already installed
# See: https://cli.github.com/

# Authenticate with GitHub
gh auth login

# Run the setup script
./scripts/setup-github-secrets.sh

Step 4: Verify Secrets

After adding all secrets:

The GitHub Actions workflow will automatically use them during deployment
Check deployment logs to ensure no authentication errors
Visit your deployed application to verify it's working

Security Best Practices

Never commit credentials to version control
Rotate API keys regularly
Use least-privilege access for service accounts
Monitor secret usage in logs and audit trails
Remove unused secrets promptly

Troubleshooting Deployment

If deployment fails with authentication errors:

Verify all required secrets are added to GitHub
Check secret names match exactly (case-sensitive)
Ensure API keys are valid and not expired
Check base URLs are accessible from your deployment environment
Review GitHub Actions logs for specific error messages

Deploy to OpenShift

Once GitHub Secrets are configured, simply push to the main branch:

git add .
git commit -m "Configure deployment"
git push origin main

The GitHub Actions workflow will automatically:

Build container images
Push to your container registry
Deploy to OpenShift
Set up networking and routes

🐛 Troubleshooting

Common Issues

"No MaaS models configured"

Check that you have at least one complete MaaS model configuration (API_KEY, BASE_URL, MODEL_NAME)
Verify environment variables are loaded: echo $MAAS_PHI_4_API_KEY

"JIRA connection failed"

Verify JIRA_URL and JIRA_PERSONAL_TOKEN are correct
Test JIRA token: https://issues.redhat.com/rest/api/2/myself
Check token permissions in JIRA settings

"Module not found" errors

Ensure you're in the virtual environment: source venv/bin/activate
Reinstall dependencies: pip install -r config/requirements.txt
Try using the automated setup: ./utils/local-app.sh

FAISS installation issues

Error: Microsoft Visual C++ 14.0 is required (Windows): Install Visual Studio Build Tools
Error: swig executable not found (All platforms): Install SWIG first
- macOS: brew install swig
- Ubuntu/Debian: sudo apt-get install swig build-essential
- RHEL/CentOS: sudo yum install swig gcc-c++ python3-devel

GitHub Actions Deployment Failures

Verify all required GitHub Secrets are set (see Security Setup section)
Check GitHub Actions logs for specific error messages
Ensure secret names match exactly (case-sensitive)
Verify API keys are valid and not expired

🔐 Security Notes

For Production Deployment

Never commit credentials to version control
Use GitHub Secrets for all sensitive values in CI/CD
Rotate API keys regularly
Use least-privilege access for service accounts
Monitor secret usage in logs and audit trails

For Development

Keep credentials in .env file (gitignored)
Use development-specific API keys when possible
Regularly update dependencies for security patches

🤝 Common Use Cases

Create an RFE: Use the chat interface to describe your enhancement request
Validate RFE Format: Check if your RFE follows proper guidelines
Search Documentation: Find relevant Red Hat AI documentation
JIRA Integration: Create and manage JIRA tickets directly
API Automation: Integrate RFE creation into your development workflow
MCP Integration: Connect AI agents (Claude Desktop, VS Code) for automated workflows

🔐 Authentication

Web Interface

Streamlit App: No authentication required - visit the web interface directly
Interactive Testing: Use the web interface for easy testing without tokens

API Access

Production API: Requires authentication token in header: Authorization: Bearer pmbot-production-token
Local Development: May not require authentication depending on configuration
Token Format: Use Bearer followed by your token

Getting Tokens

Contact: Repository maintainer for production API tokens
Local Setup: Configure your own tokens in .env file for local development
JIRA Integration: Requires separate JIRA personal access token

Ready to get started? Try the live demo or explore the API! 🚀

Name		Name	Last commit message	Last commit date
Latest commit History 1 Commit
.github/workflows		.github/workflows
.streamlit		.streamlit
config		config
deployment		deployment
docs		docs
document_cache		document_cache
documents		documents
scripts		scripts
utils		utils
vector_db		vector_db
.env.template		.env.template
.gitattributes		.gitattributes
.gitignore		.gitignore
README.md		README.md
api_server.py		api_server.py
atlassian_client.py		atlassian_client.py
auth.py		auth.py
deploy-to-cluster.sh		deploy-to-cluster.sh
document_processor.py		document_processor.py
pm_chatbot_main.py		pm_chatbot_main.py
rfe_manager.py		rfe_manager.py
vector_database.py		vector_database.py

jhuang2026/PMBot

Folders and files

Latest commit

History

Repository files navigation

PMBot - Product Manager Assistant 🤖

🌐 Live Demo

✨ Key Features

🚀 Getting Started

Option 1: Use the Web Interface (Recommended)

Option 2: Use the REST API

Option 3: MCP Integration (For AI Agents & Tools)

Quick MCP Setup

Available MCP Tools

MCP Usage Examples

🛠️ API Usage Examples

Generate an RFE

Search Documentation

🏗️ Architecture

📁 Project Structure

🔧 Local Development Setup

Prerequisites

System Dependencies

🚀 Quick Start (5 minutes - Recommended!)

Step 1: Clone & Setup

Step 2: Configure Environment

Step 3: Get Your Credentials

Step 4: Run the Application

Step 5: (Optional) Start API Server

Environment Variables Quick Reference

Required Variables

🚀 Production Deployment

Manual Deployment (Foundation)

Prerequisites for Deployment

Container Registry Setup

Using the Deploy Script

GitHub Actions Deployment (Automated)

🔐 Security Setup for GitHub Deployment

⚠️ CRITICAL: Setting Up GitHub Secrets

Step 1: Navigate to Repository Secrets

Step 2: Add Required Secrets

JIRA Configuration

MaaS Model Configuration

Step 3: Automated Setup (Recommended)

Step 4: Verify Secrets

Security Best Practices

Troubleshooting Deployment

Deploy to OpenShift

🐛 Troubleshooting

Common Issues

🔐 Security Notes

For Production Deployment

For Development

🤝 Common Use Cases

🔐 Authentication

Web Interface

API Access

Getting Tokens

About

Resources

Uh oh!

Stars

Watchers

Forks

Releases

Packages 0

Uh oh!

Contributors

Uh oh!

Languages

Packages