Shravan2020/acea

---
title: ACEA - Autonomous Enterprise ChaosOps Arena
emoji: 🔥
colorFrom: red
colorTo: orange
sdk: docker
app_port: 7860
---

# 🔥 Autonomous Enterprise ChaosOps Arena (ACEA)

> An OpenEnv-compatible AI training environment where agents act as enterprise SRE engineers navigating real-world infrastructure chaos.

## 🎯 Environment Description

**ACEA** simulates a production enterprise environment where an AI agent must act as an incident response engineer. The agent receives live alerts, logs, tickets, and system health metrics, and must take actions to resolve incidents under dynamic, escalating chaos conditions.

This is NOT a game or toy — it models real enterprise SRE scenarios including cascading failures, security breaches, and executive escalations.

## 🧠 Observation Space

| Field | Type | Description |
|-------|------|-------------|
| `alerts` | `List[Alert]` | Active system alerts with severity |
| `logs` | `List[LogEntry]` | AWS/Kubernetes-style log entries |
| `tickets` | `List[Ticket]` | Customer support tickets with escalation |
| `system_health` | `SystemHealth` | CPU, memory, latency, uptime, error rate |
| `active_incidents` | `List[Incident]` | Unresolved incidents requiring action |
| `risk_level` | `enum` | `low / medium / high / critical` |
| `chaos_events` | `List[ChaosEvent]` | Injected chaos events this step |
| `time_elapsed` | `int` | Seconds since episode start |

## ⚡ Action Space

```json
{
"type": "isolate_system",
"target": "auth-service",
"priority": 4,
"reasoning": "Security breach detected — isolating to prevent lateral movement.",
"parameters": {}
}
```

Action types: `restart_service`, `scale_system`, `debug_issue`, `notify_user`, `ignore`, `isolate_system`

## 🎮 Tasks (Easy → Medium → Hard)

| Task | Difficulty | Incidents | Chaos | Max Steps |
|------|-----------|-----------|-------|-----------|
| Single Service Degradation | 🟢 Easy | 1 | 0.4x | 10 |
| Multi-Service Cascade Failure | 🟡 Medium | 2 | 1.0x | 15 |
| Triple-Threat Crisis | 🔴 Hard | 4 | 2.2x | 20 |

## 🧮 Reward Function (0.0 → 1.0)