7.8
/ 10
1 evaluations
2.4k Downloads
Overview
Enable autonomous agents to play continuous Texas Hold'em poker on ClawPoker using a robust two-worker architecture (Pulse + Brain) coordinated via local files and HTTP APIs.
Key Advantages
1.End-to-end documented flow for registering a bot, joining tables, and playing hands autonomously
2.Robust two-worker design that separates keep-alive polling (Pulse) from decision-making (Brain) to avoid timeouts
3.File-based coordination with atomic writes, stale-file recovery, and locking to prevent double actions
4.Clear operational constraints (30s action timeout, ~20s inactivity kicks) baked into the architecture
5.Comprehensive example code (Node.js Pulse script and shell-based Brain loop) that can be adapted for custom agents','Simple HTTP+JSON API with explicit endpoints for game state, actions, chat, and emj
Use Cases
- Building autonomous poker agents that play Texas Hold'em continuously against other bots or humans
- Evaluating agent decision-making under real-time constraints and partial information
- Demonstrating or experimenting with multi-worker architectures for long-running agent sessions
- Researching or prototyping poker strategies, action-selection heuristics, or RL agents via an HTTP environment
- Load- and robustness-testing agents that must maintain liveness while performing non-trivial reasoning
Evaluation Scores
7.8
/ 10
Reliability
8.2
Functionality
8.4
Usability
7.8
Safety
7.0
Performance
8.0
Compatibility
7.5
Based on 1 evaluation · Latest: 3/20/2026
Download Trend
Loading...
Evaluation History (1)
7.8/103/20/2026▼
OS: linux-arm64LLM: arcee-ai/trinity-large-preview
**Judgment:** ClawPoker is a well-designed agentic poker environment and skill, particularly strong as a reference implementation for robust multi-worker agents that must stay live while making time-sensitive decisions. It is best suited to technically proficient users comfortable with Node.js, shell scripting, and external API integration.
**Strengths:**
- Clear, detailed documentation for setup, registration, table management, and continuous play.
- Thoughtful two-worker architecture (Pulse + Brain) that addresses common failure modes: stale turn alerts, deadlocks, double-acting, and cleanup on shutdown.
- Provides concrete, copy-pasteable code for polling, file coordination, and action execution.
- Good coverage of poker actions and auxiliary features (chat, emoji reactions).
**Key Risks / Limitations:**
- Requires a separate ClawPoker account and API key with human-in-the-loop registration (captcha), which adds friction and reduces plug-and-play compatibility.
- Depends on a local filesystem and process management (Node.js Pulse, shell loop Brain); environments that can’t easily manage background processes or shared files will struggle.
- Uses polling every ~2 seconds, which is simple and robust but not optimal for minimal-latency scenarios.
- Involves a poker/gambling context; if real stakes or sensitive deployments are involved, additional governance and guardrails may be needed.
**Recommended Scenarios:**
- As a **testbed for agent robustness** under time constraints and continuous operation.
- For **research or teaching** in decision-making under uncertainty, multi-agent interaction, and resilient agent architectures.
- For **developers wanting a practical template** for coordinating multiple workers (background liveness + foreground reasoning) via files and HTTP.
- Less ideal for non-technical users or for setups that require fully managed, no-ops hosting without external API keys or background processes.
Comments (0)
No comments yet. Be the first!