The Mavericks' Incident Handler

A suite of agentic workflows to help resolve incident reports and execute solutions

Quick Start Guide

Running the Incident Analyzer

To get started with the Incident Analyzer Flask (frontend) application:

Step 0: Install necessary packages

pip install -r requirements.txt
python -m spacy download en_core_web_sm
python -m spacy download en_core_web_md

Step 1: Navigate to the Application Directory

cd incident-analyzer-hackathon/incident-analyzer

Step 2: Run the Flask Application

Option A: Direct Python Command

python3.10 app.py

Option B: With Azure OpenAI API Key

export AZURE_OPENAI_API_KEY='your-api-key-here'
python3.10 app.py

Step 3: Access the Application

Once the server is running, open your web browser and navigate to:

http://localhost:5000

Requirements

• Python 3.10 or higher (required for fastmcp compatibility)
• Flask (will be imported from requirements if installed)
• Check the requirements.txt file for installation of required packages

Testing

To verify the application is working correctly:

cd incident-analyzer-hackathon/incident-analyzer
python3.10 test_app.py

Note About API Key

⚠️ The application will start without an Azure OpenAI API key, but incident analysis features will not work until you set the AZURE_OPENAI_API_KEY environment variable.

To set the API key:

export AZURE_OPENAI_API_KEY='your-actual-api-key'

Then start the application.

Troubleshooting

• Python version error: Ensure you're using Python 3.10+

  python3.10 --version

• Import errors: The orchestration agent integration may show warnings if dependencies are missing, but the app will still work using fallback prompts.

• Port already in use: If port 5000 is already in use, modify the port in app.py:

  app.run(debug=True, host="0.0.0.0", port=5001)  # Change to different port

Future implementation

• Voice input of incident reports to make it truly multimodal
• Using Transformer-based Spatio-temporal attention (TrAIsformer) for hetergenous inputs
• Create a sandboxed representation of knowledge
• Finetune LLMs for agent calls using Gradient Low Rank Projection (GaLore)

Created by Abhay Ganti, Ambar, Praneeth Suresh