$ show LYiHub/OpenClaw-AWD-Arena
OpenClaw AWD Arena is a comprehensive platform for conducting Attack-With-Defense (AWD) cybersecurity competitions, specifically designed for training and evaluating large language models (LLMs) in offensive and defensive security scenarios. The system orchestrates isolated Docker container environments where each participant (LLM agent or human) runs a vulnerable service that must be defended while attempting to capture flags from opponents' services. It features a web-based interface for match configuration and real-time monitoring, a Python/FastAPI backend handling game logic, scoring, and WebSocket-based event streaming, and custom orchestration logic for managing the lifecycle of match environments. The platform includes built-in flag management, scoring systems, match replay functionality, and support for multiple agent backends via a plugin architecture.
The LLM identified 5 files worth highlighting for reuse:
Defines the abstract interface for agent backends, enabling pluggable support for different LLM agent frameworks (e.g., OpenClaw, Hermes) without modifying core game logic. Demonstrates clean separation of concerns and extensibility.
Implements a dynamic registry to discover, load, and manage backend implementations by name. Allows zero-configuration extension with new agent types, showcasing a robust plugin architecture.
A self-contained React component that visualizes the network topology of ongoing matches using SVG or canvas. Isolated from application state, receives data via props, and can be reused in other monitoring dashboards with minimal adaptation.
Manages the complete lifecycle of Docker networks and containers for isolated match environments, including setup, configuration, resource allocation, and cleanup. Encapsulates complex Docker SDK interactions into a clean, reusable API for ephemeral environment provisioning.
Handles the full lifecycle of competition flags: secure generation, injection into target services, validation of submissions, duplicate prevention, and scoring logic. Demonstrates encapsulation of security-critical operations with clear separation from game state management.