dsh-holdem
Run the following command in DeepSeek Harness:
dsh plugin install BubblePtr/dsh-holdt
Paste the following prompt into your AI chat to install this plugin:
Install the dsh-holdem plugin in DeepSeek Harness to start a No-Limit Texas Hold'em game against 5 LLM agents; run `dsh plugin install BubblePtr/dsh-holdt`, and visit https://github.com/BubblePtr/dsh-holdt for the source code. Source: https://github.com/BubblePtr/dsh-holdem
About this plugin
Unleash high-tension brain games within the DeepSeek Harness Web interface. This plugin constructs a No-Limit Texas Hold'em simulation environment where you compete as a single player against five LLM agents. It is not just a game but an excellent platform to observe decision-making logic and psychological interplay under conditions of information asymmetry.
The core highlight lies in its meticulously designed interaction mechanism. Agents are opaque to each other and can only see their own hole cards, while table banter is restricted to Chinese and never leaks hand information or reasoning processes. This design perfectly replicates the noise environment of a real game, forcing agents to use speech to deceive opponents and analyze the situation, greatly enriching the layering of the confrontation.
Whether you are an AI researcher evaluating a model's game theory capabilities or a developer exploring the boundaries of multi-agent collaboration and confrontation, dsh-holdem offers intuitive and interesting feedback. By seamlessly integrating complex game logic with the Web frontend, it allows you to witness directly in the browser how agents progress towards victory or failure.
Screenshots
Use Cases
- Test DeepSeek models' decision capabilities in complex game environments
- Observe multi-agent performance under information asymmetry
- Experience web-based Texas Hold'em battles without network requirements
Best For
- AI Researchers
- Model Training Engineers
- Game Development Enthusiasts
Related Plugins
A method pack that makes AI coding agents plan against your real baseline, prove completion with fresh evidence, and reduce reworks and unsafe changes.
Turns the DeepSeek Harness session into a captain that builds a durable sub-agent team, splits goals into dependency-aware tasks, and coordinates work via direct messages and a live Web UI.
Gives coding agents design judgment, letting Claude Code, Cursor, and 70+ agents generate and iterate high-quality UI, presentations, and graphics right from the terminal.
Run the Pi ecosystem's plugins on DeepSeek Harness, unmodified, via a compatibility layer that implements Pi's public extension ABI on DSH's native services.