vaibhav2108/Autonomous_security_operations

GenAI SecOps Demonstration Platform 🛡️

A simulated, enterprise-scale Cybersecurity Operations platform designed to demonstrate how Generative AI agents can integrate with and improve security operations.

This application generates a synthetic, deterministic enterprise database containing thousands of assets, alerts, and incidents entirely in-memory at runtime. It then calculates real-time operational KPIs based on this data.

Finally, it provides interfaces to invoke specialized LangChain AI Agents (like Incident Triage, Root Cause Analysis, and Security Copilot) which use the synthesized metrics as context to generate simulated analysis, applying input/output guardrails to ensure safety.

🚀 Features

• —Dynamic Data Generation: Generates 10k+ assets and 20k+ alerts in seconds.
• —Deterministic KPI Engine: Computes MTTD, MTTR, False Positive rates natively via Pandas.
• —Interactive Dashboard: Built on Streamlit and Altair.
• —AI Agent Interventions: LangChain-powered simulations with guardrails protecting against prompt injection.

💻 Local Development

1. 1.Clone the repository.
2. 2.Set up a virtual environment and use uv/pip to install dependencies.

   pip install -r requirements.txt

1. 1.Configure Environment Variables. Copy .env.example to .env and add your OpenAI API key:

   OPENAI_API_KEY=sk-your-key-here
   MODEL_NAME=gpt-4o-mini

1. 1.Run the Streamlit app.

   streamlit run app/main.py

☁️ Deploying to HuggingFace Spaces

This repository is structured specifically to support one-click deployment to HuggingFace Spaces using the Streamlit SDK.

1. 1.Create a new Space on Hugging Face and select Streamlit as the Space SDK.
2. 2.Push this repository's contents to the Space. HuggingFace will automatically read the requirements.txt and install the necessary Python dependencies.
3. 3.Important: In your Hugging Face Space settings, navigate to the Variables and secrets section and securely add your OPENAI_API_KEY. The application will fail to run the AI agents without it.
4. 4.(Optional) Set MODEL_NAME to a different compatible model if desired.

Note: This platform is for demonstration purposes only. It does not connect to actual enterprise environments or ingest real telemetry data.