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)--- title: ACEA——自治企业混沌运维竞技场（Autonomous Enterprise ChaosOps Arena） emoji: 🔥 colorFrom: red colorTo: orange sdk: docker app_port: 7860 --- # 🔥 自治企业混沌运维竞技场（ACEA） > 一款兼容OpenEnv的AI训练环境，智能体将扮演企业站点可靠性工程师（Site Reliability Engineer，SRE），应对真实的基础设施混沌场景。 ## 🎯 环境描述 **ACEA** 模拟真实企业生产环境，智能体需扮演事件响应工程师。智能体将接收实时告警、系统日志、客户工单与系统健康指标，并需在动态演进且逐步升级的混沌条件下采取行动解决故障。 本环境并非游戏或演示工具——它精准建模了真实企业SRE场景，包括级联故障、安全入侵以及管理层升级响应流程。 ## 🧠 观测空间 | 字段 | 类型 | 描述 | |-------|------|-------------| | `alerts` | `List[Alert]` | 带严重级别的活跃系统告警 | | `logs` | `List[LogEntry]` | AWS/Kubernetes 风格的日志条目 | | `tickets` | `List[Ticket]` | 带升级流程的客户支持工单 | | `system_health` | `SystemHealth` | 涵盖CPU、内存、延迟、运行时长、错误率的系统健康状态 | | `active_incidents` | `List[Incident]` | 待处理的未解决事件 | | `risk_level` | `enum` | 可选值：`low / medium / high / critical`（低/中/高/临界） | | `chaos_events` | `List[ChaosEvent]` | 当前步骤注入的混沌事件 | | `time_elapsed` | `int` | 自训练回合（episode）开始以来的秒数 | ## ⚡ 动作空间 json { "type": "isolate_system", "target": "auth-service", "priority": 4, "reasoning": "检测到安全入侵——执行隔离以防止横向移动。", "parameters": {} } 支持的动作类型包括：`restart_service`（重启服务）、`scale_system`（扩容系统）、`debug_issue`（问题排查）、`notify_user`（通知用户）、`ignore`（忽略）、`isolate_system`（隔离系统） ## 🎮 任务（简单→中等→困难） | 任务 | 难度 | 事件数 | 混沌倍率 | 最大步数 | |------|-----------|-----------|-------|-----------| | 单服务性能退化 | 🟢 简单 | 1 | 0.4倍 | 10 | | 多服务级联故障 | 🟡 中等 | 2 | 1.0倍 | 15 | | 三重威胁危机 | 🔴 困难 | 4 | 2.2倍 | 20 | ## 🧮 奖励函数（0.0 → 1.0）