Prompt Paws: A Cozy 60 FPS AI Agent Bakery Game Teaching RAG, Guardrails & Context Caching

Description and use

Prompt Paws is a 60 FPS simulation game that teaches developers key AI engineering concepts such as guardrails, context caching, RAG, tool calling, and multi‑agent workflows.

Problem

Engineering teams need an engaging, interactive way to train developers on production LLM architecture, guardrails, caching, RAG, tool calling, and multi‑agent workflows.

Outcome

Prompt Paws provides an interactive sandbox that visualizes key AI engineering concepts, enabling teams to practice guardrails, caching, RAG, tool calling, and multi‑agent workflows in a playful environment.

Workflow

• Launch the browser game at https://robinamirbahar.github.io/cozy-ai-agent-cafe-game/
• Select a 7‑shift level to start a curriculum
• Place and upgrade Robot‑Cat Chef stations on the bakery ribbon
• Observe how each station processes orders according to its AI pattern
• Use the Auto‑Chef Blueprint button to auto‑deploy optimal pipelines
• Deploy the stateless container to Google Cloud Run with one click

Evidence

Prompt Paws: The Cozy AI Agent Cafe & Bakery – A cute 60FPS kawaii simulation game teaching RAG, Guardrails, Context Caching, Tool Calling & Multi-Agent Workflows! @google-deepmind

Source

Play Live in Your Browser (60 FPS HTML5 Canvas): https://robinamirbahar.github.io/cozy-ai-agent-cafe-game/

Source