🛠️ AI assisted DevOps Troubleshooting

An autonomous multi-agent system built with LangGraph to analyze infrastructure logs, investigate root causes via web search, and generate production-ready remediation plans.

🌟 Introduction

This project uses stateful graph architecture. By using LangGraph, the system maintains a structured state across three specialized agents:

1. Log Analyzer: Extracts failure patterns from raw logs.
2. Issue Investigator: Research solutions using Tavily Search.
3. Solution Specialist: Drafts actionable, step-by-step runbooks.

---

🚀 Setup

This project uses uv for lightning-fast dependency management.

1. Install uv (if you haven't):

   curl -LsSf https://astral.sh/uv/install.sh | sh

2. Clone & Enter the repo:

   git clone <your-repo-url>
   cd devops-ai-graph

3. Sync dependencies:

   uv sync

4. Run the application:

   uv run main.py

---

📦 Dependencies

The core stack consists of:

• LangGraph: Orchestration of agent states.
• LangChain Azure OpenAI: LLM integration for enterprise-grade reasoning.
• Tavily: AI-optimized search engine for technical documentation.
• Python-dotenv: Environment variable management.

---

🏗️ Project Structure

.
├── task_outputs/       # Generated remediation plans (.md)
├── main.py             # Entry point & Dotenv initialization
├── graph.py            # LangGraph workflow definition
├── nodes.py            # Agent logic & Prompts
├── tools.py            # File system & Search tools
├── state.py            # TypedDict state definition
├── .env                # Secrets (See below)
└── pyproject.toml      # Project config (uv)

---

Certainly! Updating the README to include descriptions for each variable makes it much more user-friendly, especially for team members who might be setting this up for the first time.

Here is the updated section for your README.md:

---

🔐 Environment Variables

The application requires several environment variables to function. You can find a template in .env.example. Create a .env file in the root directory and populate it with your credentials:

# --- Azure OpenAI Configuration ---
# Your secret API key found in the Azure Portal
AZURE_OPENAI_API_KEY=your_api_key_here

# The base URL for your Azure OpenAI resource (e.g., https://name.openai.azure.com/)
AZURE_OPENAI_ENDPOINT=https://your-resource.openai.azure.com/

# The custom name you gave your model deployment in Azure AI Studio
AZURE_OPENAI_DEPLOYMENT=your_deployment_name

# The specific API version to use (e.g., 2023-05-15 or 2024-02-15-preview)
AZURE_OPENAI_API_VERSION=2024-02-15-preview

# --- Search Tool ---
# API key for Tavily to allow the agent to perform web research
TAVILY_API_KEY=tvly-xxxx...

---

🛠️ Best Practices Included

• Path Security: ReadFileTool is restricted to the project root_dir to prevent unauthorized file access.
• Decoupled Logic: Nodes, tools, and state are separated for easy unit testing.
• Concise Prompting: Agents use structured operational prompts or better performance with GPT-4o.
• Automated Workspace: main.py automatically ensures the task_outputs/ directory exists before execution.