Simulate a tiny AI civilisation for $0.02

Goal: Build a small agent-based simulation where models choose and code enforces.
Setup: agents used in source: Devin; tools/MCPs stated: Three.js, OpenRouter.
Steps:
1. Give each citizen a role, skills, needs and memories.
2. A decision model (Jev) gets each citizen’s state, observations and allowed actions, and picks the next action.
3. A larger model acts as world director, proposing events (construction, weather, arrivals).
4. Use a coding agent (Devin) to build the simulation and its Three.js world.
5. Enforce rules in code: reject invalid actions; let local rules keep citizens moving while decisions are pending.
Rules: ask me before anything irreversible (sending, posting, paying, deleting); don’t invent data, results or tools I haven’t given you; tell me what you couldn’t verify.
Based on: Lovepreet Singh — https://x.com/SinghDevHub/status/2102086884998664521

Steps are from the source. Includes a “Based on” credit line.