📊 Performance Metrics
0.85
Task 1 Score
(Easy)
0.85
Task 2 Score
(Medium)
1.00
Task 3 Score
(Hard)
0.90
Average Score
⭐
🎯 Environment Overview
OpenOps simulates realistic production incidents where AI agents must investigate alerts, identify root causes, execute mitigation actions, communicate with stakeholders, and resolve incidents to minimize revenue loss.
✨ Key Features
🔍 Smart Detection
Intelligent incident classification based on alerts and logs
📈 Progressive Difficulty
Three tasks from simple crashes to cascading failures
⚡ Rule-Based Agent
Reliable baseline with optimized playbooks
🎮 21 Actions
Investigate, mitigate, communicate, and resolve
🔌 API Endpoints
GET /docs
Interactive Swagger UI with all endpoints
POST /reset?task_id=1
Initialize a new incident scenario (task_id: 1, 2, or 3)
POST /step
Take an action and receive observation
GET /state
Retrieve current environment state
GET /actions
Get all available actions
📖 Three Progressive Tasks
API service down due to OOM → Inspect logs → Restart → Resolve
Database deployment broke queries → Rollback → Notify team → Resolve
DB overload → API timeouts → Scale DB → Restart API → Communicate → Resolve