gitwtfhub

wtf is agentcity?

yehudalevi/agentcity — explained in plain English

Analysis updated 2026-05-18

1HTMLAudience · developerComplexity · 2/5Setup · easy

TL;DR

A local, deterministic tool that turns your AI coding agent activity history into a growing pixel art city, with no accounts or network calls.

Mindmap

mindmap
  root((agentcity))
    What it does
      City from agent history
      Deterministic generation
      Founding timelapse
    Tech stack
      TypeScript
      Node.js
      Canvas renderer
    Use cases
      Activity visualization
      Team skyline
      Progress tracking
    Audience
      Developers
      AI agent users

Code map

Detail Auto

An interactive map of this repo's files and how they connect — its source is parsed live in your browser. Click Visualize to build it.

filefunction / class

Why would anyone build with this?

REASON 1

Visualize your coding agent activity over time as a growing pixel art city.

REASON 2

See which repos you've been actively shipping work to versus neglecting.

REASON 3

Replay a timelapse of months of agent work as your city builds itself.

REASON 4

Share a combined team skyline on an office screen using the optional federation hub.

What's in the stack?

TypeScriptNode.jsHTML Canvas

How it stacks up

yehudalevi/agentcityabhas9/escape-runabhishek-kumar09/mern-live-js-june-2021
Stars111
LanguageHTMLHTMLHTML
Last pushed2021-09-13
MaintenanceDormant
Setup difficultyeasyeasymoderate
Complexity2/52/52/5
Audiencedevelopergeneraldeveloper

Figures from each repo's GitHub metadata at analysis time.

How do you spin it up?

Difficulty · easy Time to first run · 5min

Wtf does this do

Agentcity turns your history of working with AI coding agents into a persistent, pixel art city that lives entirely on your own computer. Every repository you work on becomes a building, and the more shipped work you put into that repo, the more floors and upgrades that building gets, growing from a simple tent up through a hut, house, workshop, tower, and eventually a landmark. The kind of work you do, such as writing code, tests, infrastructure, or research, shapes what district and style a building takes on, and roads between projects widen the more you move between them. The city is generated deterministically from your local agent activity log plus a seed value you choose, meaning the exact same history and seed will always produce the exact same city, with no separate save file needed. Different seeds produce different landscapes such as rivers, coasts, or hills, so no two people's cities look alike even with similar work. When you first launch it, agentcity reads your existing agent history and replays it as a sixty to ninety second timelapse showing your city grow from its first building to today, so returning users do not start from an empty field. Everything runs locally by default: there are no accounts, no telemetry, and no network calls unless you explicitly export something like a snapshot image. You can pan, zoom, and rotate the isometric camera, scrub back and forth through your city's full history, and hover over any building to see its repo name, tier, and work mix. An optional federation feature lets a team opt in to share a combined skyline on something like an office screen, but even then only privacy safe summary facts are sent, never raw commands or file paths. The tool is run with a single command using npx, and there is also a demo mode with simulated activity for anyone without existing agent history yet. It is built in TypeScript for Node 20 or later with no runtime dependencies, and the entire visual renderer is one self contained HTML file using canvas drawing rather than any framework. The project describes itself as early and actively evolving, with the solo version already usable daily.

Yoink these prompts

Prompt 1
Help me run agentcity with npx to found my city from my local agent history.
Prompt 2
Explain how the deterministic city = f(events, seed) compiler works.
Prompt 3
Walk me through setting up the federation hub for my team.
Prompt 4
How do I try agentcity's demo mode before it has real activity data?

Frequently asked questions

wtf is agentcity?

A local, deterministic tool that turns your AI coding agent activity history into a growing pixel art city, with no accounts or network calls.

What language is agentcity written in?

Mainly HTML. The stack also includes TypeScript, Node.js, HTML Canvas.

How hard is agentcity to set up?

Setup difficulty is rated easy, with roughly 5min to a first successful run.

Who is agentcity for?

Mainly developer.

View the repo → Decode another repo

This repo across BitVibe Labs

Don't trust strangers blindly. Verify against the repo.