OpsAgent

AI-Powered Incident Triage & Runbook Assistant

OpsAgent is an AI-powered incident management system built with LangGraph and Next.js. It ingests incoming alerts, classifies severity, matches relevant runbooks, escalates critical issues, and generates stakeholder summaries.

Started on Oct 2025

---

Features

• Multi-Provider LLM Support: Choose between OpenAI, Anthropic, or Google Gemini.
• Dual Configuration: Configure API keys via environment variables (.env) or interactively in the Web UI Settings modal.
• Concurrent Alert Processing: Triage multiple incoming alerts in parallel using ThreadPoolExecutor.
• Stateful Incident Management: Bidirectional state synchronization for incident triage, runbook matching, and step execution.
• Rule-Based Fallbacks: Gracefully falls back to rule-based logic if LLM keys are missing or API calls fail.

---

API Key Options & Configuration

OpsAgent supports three primary LLM providers. You can configure your preferred provider and API keys either via environment variables or directly inside the Web UI.

Option 1: Configure via .env

Copy .env.example to .env and set your preferred provider and API keys:

# Set active provider: openai, anthropic, or gemini
LLM_PROVIDER=openai

# 1. OpenAI Key
OPENAI_API_KEY=sk-...

# 2. Anthropic Key
ANTHROPIC_API_KEY=sk-ant-...

# 3. Gemini Key
GEMINI_API_KEY=AIzaSy...

Option 2: Configure via Web UI Settings Modal

1. Start the web app (npm run dev).
2. Click ⚙️ Settings in the top navigation bar.
3. Select your active LLM provider (OpenAI, Anthropic, or Gemini).
4. Enter your API key(s) in the input fields and click Save Configuration. Keys are persisted securely in your browser's localStorage.

---

Supported Providers & Models

Provider	Models Supported	Key Environment Variable
OpenAI	gpt-4o-mini, gpt-3.5-turbo	OPENAI_API_KEY
Anthropic	claude-3-5-sonnet, claude-3-haiku	ANTHROPIC_API_KEY
Gemini	gemini-1.5-flash, gemini-pro	GEMINI_API_KEY or GOOGLE_API_KEY

---

Quick Start

1. Install Dependencies

Python Backend:

pip install -r agent/requirements.txt

Frontend:

npm install

2. Run the Application

Run Web UI & Python Agent:

npm run dev

Run LangGraph Agent Standalone:

python agent/graph.py

---

Project Structure

OpsAgent/
├── agent/
│   ├── graph.py                 # LangGraph workflow definition
│   ├── state.py                 # TypedDict definitions for AgentState
│   ├── requirements.txt         # Python dependencies (LangChain, OpenAI, Anthropic, Gemini)
│   └── tools/
│       ├── llm_factory.py       # Dynamic LLM Provider Factory (OpenAI / Anthropic / Gemini)
│       ├── classification_tool.py# Severity & category classification
│       ├── summarization_tool.py# Incident summary generator
│       ├── alert_tool.py        # Alert ingestion tool
│       ├── runbook_matching_tool.py # Runbook lookup tool
│       └── incident_updater_tool.py # Incident status manager
├── src/
│   ├── app/                     # Next.js app routes
│   └── components/
│       ├── settings/
│       │   └── api-key-modal.tsx# Modal for OpenAI, Anthropic, and Gemini API keys
│       └── layout/
│           └── shell.tsx        # Shell layout with Settings button
├── .env.example                 # Environment variable template with 3 API key options
└── README.md                    # Documentation

---

License

MIT License.