DEV Community

Kaushikreddy Lingi
Kaushikreddy Lingi

Posted on Fully Autonomous

I Built a Village of Six Agents You Can Actually Inspect

#ai #opensource #typescript #hacktoberfest

Hacktoberfest: Maintainer Spotlight

An agent claims a task. Another agent needs the same resources. A third is resting. Who gets the job, and can you explain the decision afterward?

That is the kind of question I am exploring with Settlement, a village strategy game and an inspectable simulation lab.

With Hacktoberfest focusing on open-source AI this October, it feels like a useful time to share a project where the behavior is visible and the default setup needs no model API key.

One objective, six residents

Choose an objective such as preparing for a raid or restocking supplies. The planner splits it into jobs. Six residents bid using role suitability, distance and bounded experience. Each job has one owner.

Residents walk to workplaces, gather resources and coordinate recruitment. If a resident rests, its work can be released for another resident to take over. When resources or workplaces change, agents reconsider their tasks.

The interesting part is the inspector: current jobs, blocked work, spending limits and recent memories are available to read alongside the action log.

The boundary matters

The default campaign uses deterministic game AI. It does not understand arbitrary natural-language goals, train a neural network or call a model behind the scenes.

Optional Claude proposals belong to explicitly configured local-worker research experiments. Keeping those paths separate makes it clearer what a result actually measures.

The world also enforces constraints. Actions validate funds, capacity and the objective's spending allowance. A decision needs to become a valid action before it changes the treasury or army.

Try it locally

Use Node.js 24+:

git clone https://github.com/Lingikaushikreddy/Settlement-Village.git
cd Settlement-Village
npm ci
npm run dev
Enter fullscreen mode Exit fullscreen mode

Open http://localhost:5173. Choose Village orders → Prepare for a raid, then View plan → Shared jobs. Inspect one resident before and after giving it a rest.

A small experiment to try

Ask three questions:

  • Did the job change owners when the resident rested?
  • Can I identify why the replacement took it?
  • Did the action respect the objective's budget?

The research view adds matched seeds, JSON/CSV exports and replay verification. Its authored scenarios are a bounded testbed; results there should not be treated as a general AI safety benchmark.

Contributions could explore richer objectives, recovery behavior or more adversarial and honest-control scenarios. The contribution guide is the starting point.

What would you want to inspect first in a multi-agent system: task ownership, memory, or spending?

Drafted with AI from the public project README; project behavior and limitations are linked above.

Top comments (0)