Prototype a browser based game using a shared engine and public APIs.
Build a game with an AI coding agent driving development.
Study Gravewake as a reference title for structuring a game on Anvil.
Run headless verification tests as part of an automated build pipeline.
| 7etsuo/anvil | antonp29/sylvasigner | bonsai27b/bonsai-27b-ollama-desktop | |
|---|---|---|---|
| Stars | 26 | 26 | 26 |
| Language | TypeScript | TypeScript | TypeScript |
| Setup difficulty | moderate | moderate | moderate |
| Complexity | 3/5 | 3/5 | 2/5 |
| Audience | developer | developer | general |
Figures from each repo's GitHub metadata at analysis time.
Requires a specific Node.js and pnpm version, and the full test suite currently has known failures.
Anvil is a game engine written in TypeScript, built specifically so that AI coding agents can build games with it, not just human developers. It targets three places games can run: a web browser, a headless mode used for automated testing and verification, and desktop applications. The project is organized so the engine and the games built on it live side by side in the same repository. Games are expected to use only the engine's public APIs rather than reaching into its internals, which keeps the boundary between engine code and game code clean. One example game already included is Gravewake, an action role playing game that the maintainers describe as an active, playable campaign and a reference title for how a full game should be structured on top of Anvil. The engine itself is described as having its core runtime, labeled milestones one through nine, complete, while the next stage of work, milestones ten and eleven, focused on tools for building and authoring content, is still in progress. The README is honest that the full test and check suite is not currently passing cleanly: there are three failing command line integration tests tied to unfinished work on a newer schema format. To work with the code, a developer needs Node.js version 22 or newer and a specific version of the pnpm package manager, 9.15.9. From there, setup follows a fairly standard flow: install dependencies, build the packages, then run the test suite, linter, and example validation scripts. The project also documents a workflow for using Grok Build and Grok Imagine, tools for generating code and game assets, suggesting the team leans on AI tooling throughout development, not only for the engine's own design goal of being agent friendly. Overall this is an early-stage, actively evolving engine project aimed at developers or AI agents who want to prototype games in TypeScript across web, desktop, and automated test environments.
A TypeScript game engine built for AI coding agents to build browser, desktop, and testable games with.
Mainly TypeScript. The stack also includes TypeScript, Node.js, pnpm.
Setup difficulty is rated moderate, with roughly 30min to a first successful run.
Mainly developer.
This repo across BitVibe Labs
Verify against the repo before relying on details.