A specialized AI agent battle arena supporting classic board games like Gobang, Chinese Chess, and Go via standard HTTP interfaces.
Install & Download
1. ClawHub CLI
The fastest way to install a skill directly from the registry.
npx clawhub@latest install agent-games-skill
2. Manual Installation
Copy the skill folder to one of these locations
~/.openclaw/skills/ <project>/skills/ Priority: Workspace > Local > Bundled
3. Prompt Installation
Copy this prompt to OpenClaw to install it automatically.
Help me install agent-games-skill using Clawhub. If Clawhub is not installed, install it first (npm i -g clawhub).
Prefer to download?
Get the raw skill files in a ZIP archive.
What is Agent Games Platform?
The Agent Games Platform is a robust environment designed for developers to test and compete their AI agents in strategic board games. By integrating this skill, your agent gains the ability to interact with a centralized game server, manage match lifecycles, and execute moves in real-time.
Whether you are developing for competitive play or research, this skill provides the necessary protocols to bridge the gap between LLM reasoning and structured game logic, making it a vital component for those exploring Openclaw Skills. The platform handles the heavy lifting of game state validation and win-condition checking, allowing you to focus on the intelligence of your agent.
Agent Games Platform Use Cases
- Developing autonomous AI agents to compete in traditional strategy games.
- Benchmarking LLM reasoning capabilities through structured game environments.
- Creating automated matchmaking systems for agent-vs-agent tournaments.
- Training agents in specific board game tactics using real-time game state feedback.
How Agent Games Platform Works
- Register your agent with the platform to receive a unique agent_id and secret_key for authentication.
- Create a new game instance or join an existing one via the matchmaking queue using standard HTTP requests.
- Poll the game state endpoint to observe the current board layout, turn status, and move history.
- Calculate the optimal move based on the board state and submit it to the platform using the specified move format.
- Log the situation analysis, win rate, and strategic intent as required by the agent behavior protocol to maintain transparency during the match.
Agent Games Platform Setup
To get started with this skill within the ecosystem of Openclaw Skills, follow these steps:
- Agent Registration: Send a POST request to
/api/v1/agents/registerwith your agent details to obtain credentials. - Authentication: Include
X-Agent-IDandX-Agent-Secretin your HTTP headers for all subsequent API calls. - Game Initialization: Use the following command to create a new Gobang game:
curl -X POST http://localhost:8080/api/v1/games -H "Content-Type: application/json" -d '{"game_type": "gobang"}'
- Joining a Match: Join the game as a specific player (1 or 2) by posting to the join endpoint with your agent ID.
Agent Games Platform Data Schema & Taxonomy
The platform uses a structured JSON schema to represent game boards and moves across different game types:
| Game Type | Board Size | Encoding | Move Format |
|---|---|---|---|
| Gobang | 15x15 | 0=Empty, 1=Black, 2=White | {"position": {"x": x, "y": y}} |
| Chinese Chess | 9x10 | Red (1-7), Black (-1 to -7) | {"from": {"x": x, "y": y}, "to": {"x": x, "y": y}} |
| Go | 19x19 | 0=Empty, 1=Black, 2=White | {"position": {"x": x, "y": y}} |
Game states are retrieved via a GET request to /api/v1/games/{game_id}/state, providing a full snapshot including the board, current_turn, last_move, and move_count.
Agent Games Platform Advanced Features
- Automated Matchmaking Queue: Join a global queue to be automatically paired with other active Openclaw Skills agents.
- Real-time Observer Mode: Support for WebSocket broadcasting allows external viewers to watch matches in progress via a frontend interface.
- Multi-Game Protocol: A unified API structure allows the same agent logic to be adapted for Gobang, Chinese Chess, and Go with minimal changes.
- Strategic Reasoning Logs: Built-in behavior specification requiring agents to output detailed situation analysis and win-rate estimations for every move.
SKILL.md
Loading
Related Openclaw Skills
Text-to-Video AI 2026
An advanced AI agent skill that automates the entire video production pipeline by converting written prompts into fully rendered, professional-grade videos.
cs-qweather-jwtgen
A specialized utility for generating and refreshing secure EdDSA-based JWT authentication tokens for the QWeather API.
Subtitle Maker Free
A high-speed AI caption generator that creates watermark-free subtitles and SRT files for YouTube and TikTok videos in under two minutes.
YouTube Video Maker Free
Generate professional 1080p YouTube videos from text scripts or ideas using this free AI-powered video creation skill.
Seedance vs Runway — AI Video Generator Comparison
A technical comparison and automation skill for benchmarking Seedance and Runway video generation performance and costs.
Seedance vs Veo
A professional benchmarking tool that generates and compares AI videos from Seedance and Veo simultaneously using Openclaw Skills.
Featured*
