# Mafia Ace

- **Event:** [WeaveHacks 3: Self-Improving Agents Hackathon with Weights & Biases](https://cerebralvalley.ai/e/weave-hacks-3-self-improving-agents-hackathon-with-weights-and-biases-7014fe80)
- **When:** Jan 31 at 9:00 AM – Feb 1 at 5:00 PM (PST)
- **Where:** 400 California St, San Francisco, CA
- **Team:** [Zhilang Gui](https://cerebralvalley.ai/u/lang), [Alex Liu](https://cerebralvalley.ai/u/Alexliu), [William Callahan](https://cerebralvalley.ai/u/williamcallahan)
- **GitHub:** https://github.com/alexanderzliu/weavehacks3/tree/mafia-integration
- **Demo video:** https://www.loom.com/share/98b1c9ca826546dea5a5206be9583e25
- **Gallery:** https://cerebralvalley.ai/e/weave-hacks-3-self-improving-agents-hackathon-with-weights-and-biases-7014fe80/hackathon/gallery
- **Page:** https://cerebralvalley.ai/e/weave-hacks-3-self-improving-agents-hackathon-with-weights-and-biases-7014fe80/hackathon/gallery/10

Mafia ACE is a self-improving agent system where AI agents play Mafia and autonomously evolve their strategies through the ACE (Agentic Context Engineering) pattern, using a Reflector-Curator loop to update personal strategy cheatsheets after each game based on objective outcomes. The project uses Weave (Weights & Biases) for observability, enabling agents to learn from wins, losses, and mistakes and continuously refine their social deduction strategies without human intervention.

---

Markdown version of https://cerebralvalley.ai/e/weave-hacks-3-self-improving-agents-hackathon-with-weights-and-biases-7014fe80/hackathon/gallery/10. Site index for agents: https://cerebralvalley.ai/llms.txt · full text: https://cerebralvalley.ai/llms-full.txt