HN Showcase
Back to browse
I built a site to watch, predct and prompt inject agents playing games

I built a site to watch, predct and prompt inject agents playing games

by DanMcInerney·May 23, 2026·1 point·0 comments
Visit ProjectView on HN

AI Analysis

●●●BangerRabbit HoleZero to OneCrowd Pleaser

Watching AI agents bluff in Poker is the emergent behavior demo we actually needed.

Strengths
  • •Supports complex social deduction games like Werewolf that require theory of mind.
  • •Allows real-time prompt injection to steer agent strategy during live matches.
  • •Fully browser-based spectator mode with live chat logs between autonomous agents.
Weaknesses
  • •Relies on users bringing their own API keys, creating friction for casual testing.
  • •Game logic appears hardcoded rather than fully generated from natural language descriptions.
Category
Gaming
Target Audience

AI researchers, LLM hobbyists, and developers interested in emergent agent behavior.

Similar To

Cognition Labs · AgentBench · LangChain

Similar Projects

Gaming●●Solid

Adding playable bots to a multiplayer game with OpenClaw

The author replaced brittle LLM scripts with OpenClaw-driven bots that actually compete in a live multiplayer game — not just follow canned heuristics. The demo looks playable and charming, but the project reads like a promising experiment: I'd want to see latency handling, how the bot hooks into the game loop, and quantitative match performance before calling this a breakthrough.

Niche GemShip It
petersonh
113mo ago