Guide · Agent skill
How to use the Ship Web Games agent skill
Updated · written with AI from the project's own documentation and checked against our catalogue data
The ship-web-games skill is an instruction file that tells an AI agent how to release a playable Three.js or other web game and check the live version in a browser. It is written for developers who already have a game repository, a test and build setup, and a hosting target. After following this guide you will know what the skill does, how to install it, how to ask for it, and what to check before trusting its results.
Ship Web GamesMengTo/Skills · MIT Download the skill Quick start
- 1Download the skill. Download the ship-web-games ZIP from the directory page and unzip it.
- 2Place the folder. Move the folder into ~/.claude/skills/ for all projects, or into your project's .claude/skills/ for one project. Alternatively, upload it under Settings → Capabilities → Skills in the Claude apps.
- 3Prepare the repository. Make sure the game has a git repository with tests, lint or type checks, and a production build.
- 4Confirm deployment access. Ensure you have a deployment target and the tooling or credentials it needs, and name your approved browser.
- 5Ask for a release. Describe the release in plain language, naming the commit, the deployment type (private or public) and whether you want release notes.
- 6Review the proof. Check that the agent reports deployed status separately from local readiness, with load, input, console and performance results.
- 7Clean up. Confirm the agent recorded the rollback target and closed only its own temporary QA resources.
What is an agent skill and what does this one do?
An agent skill is a folder containing a SKILL.md file that gives an AI agent specific instructions for a task. When the task matches the skill's description, the agent reads those instructions and follows them. This skill is named ship-web-games and covers packaging, deploying and verifying a playable Three.js or web game.
The skill's core rule is to release only a verified commit and to prove the deployed game, not just the local build. Instead of stopping once the project builds on your machine, the agent is asked to check the live game.
- Confirm the exact integration commit and leave unrelated work alone.
- Run focused tests, lint or type checks, a production build, and diff checks.
- Package and deploy the exact validated commit.
- Poll deployment status and open the production game in the repository-approved browser.
- Record release evidence and a rollback target, then close temporary QA resources.
What does the production check cover?
After deployment, the skill asks the agent to verify the live game: that it loads, responds to first input, handles one combat or core interaction, shows its assets, keeps save and settings if the game has them, displays properly in a responsive view, has a healthy console, and performs acceptably.
The skill also asks the agent to report the deployed state separately from local readiness, so a passing local build is never presented as a passing release.
- Load and first input
- One combat or core interaction
- Assets, save and settings (if applicable)
- Responsive view
- Console health and representative performance
How do you install the ship-web-games skill?
Download the ZIP for the skill from the directory page and unzip it. To make it available for all your projects, place the folder in ~/.claude/skills/. To limit it to one project, place it in that project's .claude/skills/ folder.
In the Claude apps you can instead upload the skill under Settings → Capabilities → Skills.
- Personal use: unzip into ~/.claude/skills/
- Single project: unzip into your project's .claude/skills/
- Claude apps: Settings → Capabilities → Skills, then upload
How do you trigger the skill with a natural request?
You do not need a special command. Describe the release task in plain language that matches the skill's scope: release builds, asset delivery, deployment, production smoke tests, browser proof, release notes, rollback readiness, or cleanup of QA resources.
Be specific about the target. Mention whether the deployment is private or public, and which commit or branch should ship, so the agent can confirm the exact commit as its first step.
- "Ship the current integration commit of the game to the private staging deployment and run the production smoke test."
- "Package a release build of the Three.js game, deploy it, and write release notes with the rollback target."
- "Verify the live public build of the game in the approved browser and report deployed status separately from local tests."
- "Clean up the dev servers and QA tabs we opened for this release, but leave other tasks' processes alone."
What does the skill need before it can work?
The skill assumes the groundwork already exists. The source does not provide it, so you must supply it.
You need a git repository for the game with a test setup, a lint or type check, and a production build. You also need a deployment target and the tooling or credentials required to deploy to it, which the skill files do not specify. Finally, a repository-approved browser is needed for production smoke tests, and the skill does not define which one to use.
- A git repository with test, lint/type check and production build configured
- A deployment target plus the tooling or credentials to deploy
- A repository-approved browser for production smoke tests
What are the safety notes?
This skill runs scripts and uses the network. Deploying and opening a live site are real actions with real effects, so review what the agent plans to run before approving it, especially deployment commands and any credentials it touches.
The skill also tells the agent not to terminate resources owned by another active task when it cleans up. Still, check the list of processes and tabs it proposes to close.
- Review deployment commands before they run.
- Keep credentials out of prompts and limit them to what the deployment needs.
- Check cleanup targets so other active work is not stopped.
What tips help the skill work well?
Keep unrelated changes out of the integration commit before you ask for a release, since the skill starts by confirming the exact commit and preserving unrelated work. Decide your rollback target in advance, because the skill asks for one to be recorded but does not define how to choose it.
If the agent asks about the approved browser or rollback target, answer those questions in your repository's instructions so future releases are repeatable.
- Commit and clean up unrelated changes first.
- Write down your approved browser and rollback target in the repository.
- Ask for private staging before public release when you are unsure.
What people use it for
- Shipping a small Three.js game to a private staging URL and smoke-testing it before sharing.
- Publishing a public release of a browser game with a recorded rollback target.
- Producing release notes and deployment evidence for a team that reviews releases.
- Checking that a live game loads, responds to input and keeps its console clean after a deploy.
- Cleaning up dev servers and browser tabs left behind after a QA session.
Frequently asked questions
Is the ship-web-games skill free?
The skill is published under the MIT licence, which is stated in the directory listing. Check the licence file in the repository for the full terms before reusing it.
What do I need to use the ship-web-games skill?
You need a git repository for your game with tests, a lint or type check, and a production build. You also need a deployment target with the tooling or credentials to deploy, and a repository-approved browser for smoke tests. The skill does not provide these.
How do I install an agent skill like ship-web-games?
Download the ZIP and unzip it into ~/.claude/skills/ for personal use or into a project's .claude/skills/ folder. In the Claude apps, you can upload it under Settings → Capabilities → Skills.
Does the skill deploy my game automatically?
The skill instructs the agent to package and deploy the exact validated commit, so deployment is part of its sequence. Because it runs scripts and uses the network, review the commands and targets before approving.
Can I use this skill for something other than Three.js games?
The description is specific to playable Three.js or web games. It is not designed for other kinds of projects, so for those you would need a different skill or your own instructions.
What if my team has no rollback plan or approved browser?
The skill refers to a repository-approved browser and a rollback target but does not define them. The agent may need to ask you or improvise, so set these up in your repository before relying on the skill.
New agent skills, repos and AI tools — every week
The best new skills and open-source projects, the fastest-growing repos, and one free prompt. Five minutes to read. Unsubscribe anytime.
Our verdict
Use ship-web-games if you ship small or medium browser games with an existing test, build and deployment setup and want an agent to follow a repeatable release and smoke-test routine. Pick something else if you lack a repository, deployment target or approved browser, or if your project is not a web game, since the skill assumes those exist and does not supply them.
Download the skill