Claude Code game development: a practical guide

What Claude Code does for a game

Claude Code is Anthropic’s coding agent. It reads your project, edits files and runs commands, in a terminal, in VS Code and JetBrains, in a desktop app or on the web, according to Anthropic’s documentation. Most of those surfaces need a Claude subscription or an Anthropic Console account. Tool facts on this page are as of September 2026 and link to their sources.

You still choose the engine. With a web library such as Three.js, Claude Code writes the game code and can run and check it in Chrome itself. With Godot or Unity it writes GDScript or C# like any other code, and an MCP server, added with claude mcp add (MCP docs), lets it work the editor as well:

• Godot. The community godot-mcp server (MIT) launches the editor, runs the project and reads its debug output. Add it with claude mcp add godot -- npx @coding-solo/godot-mcp.
• Unity. The Unity CLI replaces the MCP server Unity deprecated in its AI Assistant package. Run unity mcp configure claude to connect Claude Code to a Unity 6.0 LTS or later editor. The community MCP for Unity server (MIT) covers Unity 2021.3 LTS through 6.x.

MCP servers for game development covers setup and the safety rules for both.

Set up a game folder

Start with an empty folder, initialize git, open a terminal there and run claude. Anthropic’s docs list a native installer, Homebrew and WinGet, and Claude Code asks you to log in the first time.

Then add a CLAUDE.md file. Claude Code reads it at the start of every session, and /init drafts one from your project (best practices). Keep it short, because Anthropic warns that a bloated file makes Claude ignore your actual instructions. For a web game, four lines cover most of it:

- Stack: Three.js from a CDN import map, one index.html, modules in src/
- Run: npx serve . then open http://localhost:3000
- Controls: keyboard and touch, every action on both
- Rules: relative asset paths only; the game stays playable after every change

The Three.js manual documents that no-build setup, where an import map loads the library from a CDN (Three.js manual). Larger projects install it from npm with a build tool instead.

Write a first prompt that fits one small game

Give Claude Code a small, checkable target. Anthropic’s guide says precise prompts save corrections, and for larger features it suggests letting Claude interview you and write a spec file before any code (best practices). A first prompt in that spirit:

Interview me about a small browser game, then write the answers to SPEC.md. The game runs in one page with Three.js, works with keyboard and touch, and loads in seconds with no login. It is done when I can play a 60-second round on my phone.

Once SPEC.md exists, start a fresh session to build from it. Anthropic recommends the switch because the new session’s context holds the spec, without the whole interview. The no-login, instant-load constraints mirror the Vibe Jam 2026 rules (jam page).

Settle phones and multiplayer in the spec too. Among the 169 listed Claude Code games, 111 are flagged as playable on mobile and 77 have online multiplayer.

Build one mechanic per loop

Work in loops of one feature each. Anthropic’s guide describes four phases per loop, explore, plan, implement and commit (best practices).

For a change that touches several files, start in plan mode. Press Shift+Tab until plan mode shows, or start with claude --permission-mode plan, and Claude reads and proposes before it edits. Anthropic’s rule of thumb is to skip the plan when you could describe the diff in one sentence.

Add mechanics in order (movement, then hazards, then scoring, then polish), and commit whenever a loop leaves the game playable. /rewind restores earlier code and conversation, but it only tracks Claude’s own file edits, not changes made through shell commands, and Anthropic says it isn’t a replacement for git.

If you correct Claude twice on the same problem, run /clear and rewrite the prompt with what you learned. Anthropic says a clean session with a better prompt almost always does better than a long one full of corrections.

Test the game like a player

Give Claude a check it can run: tests, a build or a screenshot to compare. Without one, Anthropic’s guide says, “looks done” is the only signal Claude has.

For a browser game, connect Chrome with claude --chrome. Claude can then open your local server, read console errors and take screenshots (Chrome integration). Ask for the check in plain words:

Open localhost:3000, start a round, check the console for errors and save a screenshot of the result.

The integration needs Chrome, Edge or another Chromium browser, the Claude in Chrome extension and a direct Anthropic plan such as Pro or Max. It doesn’t run in WSL, and it stays off if you signed in with an API key.

For checks you can repeat after every change, playtesting games with AI adds a window.gameState() hook and a Playwright smoke test. Then play the game yourself, on a phone if you flagged mobile. Anthropic’s guide is blunt about the rest: “If you can’t verify it, don’t ship it.”

For touch controls, screen fit and testing on a real device, see how to make a mobile game with AI.

Publish the game

A single-player browser game needs only static hosting. itch.io takes a zip with an index.html and requires relative asset paths, because the game is hosted in a subdirectory of its HTML CDN. The zip may hold up to 1,000 files and 500 MB unpacked (itch.io docs). Cloudflare Pages deploys from a Git repository and serves the game on a pages.dev address (Cloudflare docs).

Many of the games we list play on vercel.app, pages.dev, netlify.app or github.io addresses. Ask Claude Code to check every asset path before you upload.

Then submit your game. We list games that are playable, substantially made with AI, built around a gameplay loop, safe, suitable for a general audience and not duplicates; the editorial policy has the details. If you plan to charge for it, read selling AI-made games.

What the catalog’s Claude Code games show

Data note: as of September 30, 2026, the catalog lists 404 games, and 367 of them are Vibe Jam 2026 entries (jam rules: at least 90% of the code written by AI). Creators name the tools themselves, mostly in the jam submission form. We don’t verify them.

By that count Claude Code is the most-named tool, on 169 of the 404 games (42%), ahead of Cursor (86) and Codex (73). The table compares those games with the full catalog.

Measure	Claude Code games	All 404 listed games
Listed games	169	404
Vibe Jam 2026 entries	158	367
Use Three.js	106	266
No engine stated	33	74
Use Phaser	8	20
Playable on mobile	111	256
Online multiplayer	77	173

A Game About Capybaras Delivering Food won Vibe Jam 2026 (results post), and its creator lists Claude Code as the tool. The jam drew 945 entries (jam press page). Of the listed games that placed in its top 25, 7 name Claude Code, including #8, #13, #15, #20, #22 and #23.

The five examples below cover an open-world delivery game where you balance a fruit stack, a stealth climber, a ragdoll brawler, a dogfighter where you draw your own plane and a kaiju roguelite. Across all 169, action is the most common genre, with 75 games.

Of those games, 158 were built for the jam, between April 1 and May 1, 2026, so expect jam-sized games. The creator of Floppy Brawler says Claude’s very first output, on day one, inspired the project, though they now think that early version looks terrible. Browse them all on the Claude Code hub or in the Vibe Jam 2026 archive.

Cost and models, as of September 2026

Claude Code is included in paid Claude plans. Pro costs $20 a month, or $17 a month billed annually, and Max starts at $100 a month (Anthropic pricing). Usage limits are shared with the Claude chat app (Claude Help Center). With an Anthropic Console account you pay API rates per token instead.

Anthropic’s model overview lists four current models and suggests starting with Claude Opus 5.5 for most workloads.

Model	Anthropic’s description	API price per million tokens (input / output)
Claude Fable 5.1	Demanding reasoning and long-horizon agentic work	$10 / $50
Claude Opus 5.5	Long-running agentic coding and knowledge work	$4 / $20
Claude Sonnet 5.5	The best combination of speed and intelligence	$2 / $10
Claude Haiku 4.5	The fastest model with near-frontier intelligence	$1 / $5

Use /model to switch models in a session. Aliases such as opus and sonnet point to the latest model in each tier, and opusplan plans with Opus, then switches to Sonnet to write the code (model configuration). Only 5 of the 169 Claude Code games name a model, most often Claude Opus 5.5, so this page cannot say which Claude model made most of them.

Keep building with Claude Code

• Claude Code skills for game development: skills, hooks and subagents for your second game.
• Best AI for game development: Claude Code next to Codex, Cursor and three other tools.
• Codex game development: the same workflow with OpenAI’s coding agent.
• Claude Code vs Codex for game development: the two agents side by side.
• Make a game with AI: 10 prompts to copy and adapt: prompts for each step of a build.
• How to make a game with AI and vibe coding a game: the tool-neutral basics.
• Hubs: games made with Claude and Three.js games.