hearth

作者 echoo19

Hearth: an open-source, agent-native desktop app for agentic game development.

100
Stars
12
Forks
TypeScript
语言
2026/8/24
添加时间

⚠️ 第三方软件声明

本 Skill 为第三方开源软件,独立托管于 GitHub。SkillTip 仅为信息目录,不控制或维护底层仓库。所显示的安全检查为自动化且范围有限,安装前请自行审查源码。

阅读服务条款

安装

添加到你的 Claude Code skills 目录:

# Add to your Claude Code skills
git clone https://github.com/echoo19/hearth

快速入门

使用 hearth 等 Skills 的指南。

安全报告

已验证

上次扫描:—

{
  "status": "PASSED",
  "issues": []
}

README.md

Hearth

The app for open-ended agentic game dev

Download · Quickstart · How it works · Feedback

License: MIT

The current release is v1.9.0, for macOS, Windows, and Linux.

The Hearth window: a sidebar of project folders on the left, a conversation with a coding agent in the middle, and the game running in a pane on the right

What it is

Type what you want to play. A coding agent builds it, and the game appears in the pane next to the conversation and keeps running while the agent works. Ask for a change and watch it land, rather than reading a summary of it.

There is no project format and no engine to learn. The agent writes plain files into a folder on your computer, using whatever it thinks fits, and anything that runs in a browser runs in the pane. Your game never has to know Hearth exists.

And the agent is whichever one you already pay for.

A playtester who remembers

Every project gets one. It plays your game itself, in its own browser, and keeps notes on what it found. Before it plays again it reads those notes back. So it does not arrive blank each time. It arrives knowing where it got stuck last week, and goes to see whether you fixed it.

That memory is what lets it answer the question you actually ask after every change, and the one nothing without a history can: did that help?

Here is a real session, the second one it played on a small platformer, after the jump height went up:

It says your last change helped.

What it thought you changed. The player's jump height was changed.

Why it says that. This time I saw the score reach 1 on picture 3, while last time I never collected any coins.

Anything worse. Nothing got worse.

Still could not work out. I could not work out how to reliably land on the first platform.

Every claim is pinned to the frame it happened on, so you can go and look. Its memory is a plain markdown file in your project, so when it gets something wrong about your game you open it and fix the line. Its old sessions never get rewritten, so you can catch it changing its mind.

It is not a good player and it will not pretend to be. Anything that wants reflexes will beat it, and what it writes down is what it tried and where it failed, never a claim that it finished. It will not tell you whether your game is fun either. That part is still yours.

It plays only when you ask, and each turn runs on your own model quota. Even when you stop it mid-session, it still writes the note. More in docs/tester.md.

Bring your own agent

Claude. Hearth runs the Claude Code you are signed into, so an API key is optional. You pick the model in the composer, and switching mid-chat takes effect on your next message. There is also a per-turn effort dial for when a change deserves more or less thought.

ChatGPT. Hearth talks to the open-source Codex CLI you installed, and Codex holds the sign-in. Models come live from the binary, reasoning effort included.

Anything else. The sidebar has a terminal beside the chat: a real shell, opened in the project folder, with the hearth tools on PATH. Run whatever agent CLI you like, and the pane keeps up because Hearth watches the folder either way.

Conversations survive, too. A chat keeps the agent's working memory across app restarts and disconnects, and stopping a turn pauses the conversation rather than ending it. Live slash commands, structured questions, exact approvals, workflows and subagents come from each installed CLI. When a feature belongs to the full-screen terminal UI, Continue in CLI resumes that exact provider session without letting the native adapter run beside it. Setup, attachments, skills, and approvals are in docs/agents.md.

Dev Team mode lets a lead plan and review while Hearth runs a small team of Claude or Codex engineers in parallel. See docs/devteam.md.

Your files, your folder, no lock-in

Every project is an ordinary folder under ~/Hearth. Move it, back it up, open it in any editor, put it on GitHub, or walk away from Hearth entirely and take the game with you. Nothing lives in a database or in the cloud, and there is nothing to export because it was never trapped anywhere.

Your conversations sit in there too, so the history of how the game got made travels with the project. Hearth's own bookkeeping stays in a .hearth folder that is kept out of your git history automatically. docs/projects-and-chats.md has the layout.

Going deeper

Contributing and license

Dev setup, ground rules, and the AI contribution policy are in CONTRIBUTING.md. Hearth is free and open source under the MIT license.

常见问题

What is hearth?

hearth is an open-source ai agents skill for AI coding assistants such as Claude Code, Codex CLI, and ChatGPT, built by echoo19. Hearth: an open-source, agent-native desktop app for agentic game development. It has 100 GitHub stars.

Is hearth safe to use?

hearth failed SkillsLLM's automated security scan, which flagged one or more high-severity issues. Review the Security Report section carefully before using it.

How do I install hearth?

Clone the repository with "git clone https://github.com/echoo19/hearth" and add it to your Claude Code skills directory (see the Installation section above).

What programming language is hearth written in?

hearth is primarily written in TypeScript. It is open-source under echoo19 on GitHub, so you can review or fork the full source.

Are there alternatives to hearth?

Yes. SkillsLLM lists many other AI Agents skills you can browse and compare side by side. Open the AI Agents category from the badge at the top of this page, or use the Related Skills and comparison links further down to weigh hearth against similar tools.

评论 (0)

暂无评论,成为第一个分享想法的人!

ECC

by affaan-m

10

The agent harness performance optimization system. Skills, instincts, memory, security, and research-first development for Claude Code, Codex, Opencode, Cursor and beyond.

242,21936,702JavaScript
AI 智能体ai-agentsanthropicclaude-code
查看详情
15

An agentic skills framework & software development methodology that works.

234,96620,863Shell
AI 智能体ai-agentsbrainstorming
查看详情

hermes-agent

by NousResearch

10

The agent that grows with you

234,43747,175Python
AI 智能体ai-agentsagent-orchestration
查看详情

The agent harness performance optimization system. Skills, instincts, memory, security, and research-first development for Claude Code, Codex, Opencode, Cursor and beyond.

185,94028,768JavaScript
AI 智能体ai-agentsanthropicclaude-code
查看详情

cc-switch

by farion1231

3

A cross-platform desktop All-in-One assistant for Claude Code, Codex, OpenCode, OpenClaw, Grok Build & Hermes Agent. Only official website: ccswitch.io

128,8688,826Rust
AI 智能体claude-codeai-tools
查看详情

claude-code

by anthropics

Claude Code is an agentic coding tool that lives in your terminal, understands your codebase, and helps you code faster by executing routine tasks, explaining complex code, and handling git workflows - all through natural language commands.

120,03119,897Shell
AI 智能体
查看详情

开发者还喜欢

基于喜欢此 Skill 的开发者投票和收藏

ECC

by affaan-m

10

The agent harness performance optimization system. Skills, instincts, memory, security, and research-first development for Claude Code, Codex, Opencode, Cursor and beyond.

242,21936,702JavaScript
AI 智能体ai-agentsanthropicclaude-code
查看详情
15

An agentic skills framework & software development methodology that works.

234,96620,863Shell
AI 智能体ai-agentsbrainstorming
查看详情

hermes-agent

by NousResearch

10

The agent that grows with you

234,43747,175Python
AI 智能体ai-agentsagent-orchestration
查看详情

n8n

by n8n-io

12

Fair-code workflow automation platform with native AI capabilities. Combine visual building with custom code, self-host or cloud, 400+ integrations.

201,88160,308TypeScript
MCP 服务器apisai-tools
查看详情

The agent harness performance optimization system. Skills, instincts, memory, security, and research-first development for Claude Code, Codex, Opencode, Cursor and beyond.

185,94028,768JavaScript
AI 智能体ai-agentsanthropicclaude-code
查看详情

cc-switch

by farion1231

3

A cross-platform desktop All-in-One assistant for Claude Code, Codex, OpenCode, OpenClaw, Grok Build & Hermes Agent. Only official website: ccswitch.io

128,8688,826Rust
AI 智能体claude-codeai-tools
查看详情