Documentation
Everything here is generated from the docs in the repository, so the site and the source never drift apart.
Get started
Quickstart
Install Hearth and build your first playtested game in ten minutes.
Desktop App
Running Hearth as a desktop app: Electron packaging and native folder dialogs.
Editor Guide
The editor’s chrome, keyboard shortcuts, and direct-manipulation scene tools: Checkpoint/Review, Undo/Redo, and resize/rotate handles.
Build
Components
All component types with their properties and defaults.
Prefabs
Reusable entity templates: tracked-stamp create/place/update/sync, CLI + MCP parity, and runtime spawning via ctx.scene.spawnPrefab.
Asset Pipeline
Importing real art, spritesheets, music, and fonts: slicing, animations, and the playtest step that proves audio happened.
Effects Guide
Camera.postEffects (bloom, CRT, vignette, chromatic aberration, pixelate, color grade) and per-sprite SpriteEffects (outline, hit flash, dissolve): data-driven, deterministic, and safe in headless mode and in exports.
Input Guide
Actions and virtual axes: keyboard and gamepad bindings, ctx.input, deadzone handling, and the editor Input panel.
UI Guide
Game UI built from ordinary components: UIElement, UILayout, UISlider, UIToggle, and keyboard/gamepad focus navigation via ctx.ui.
Script
Ship
Export
Turn a project into a static playable web build or a native desktop app: folder, single file, itch.io-ready zips, and Electron packaging for mac/Windows/Linux.
Hosting a Web Build
Put a Hearth web export on your own domain, a static host, a subfolder, or an iframe embed: HTTPS, gamepad/audio, and the single-file build for local double-clicks.
Shipping to itch.io
Uploading a Hearth web or desktop export to itch.io: the zip Hearth already writes, embed sizing, per-platform channels, and scripted pushes with butler.
Distributing a Desktop Game
Getting an exported native game to players, and honest first-launch guidance for unsigned builds: macOS Gatekeeper/Open Anyway, Windows SmartScreen, and how signing removes the warnings.
Agents
Agent Panel
The editor’s embedded terminal for your own coding agent: permission modes, the live activity timeline, Snapshot/Review/Revert, and the subscription-safety position behind it.
Agent Workflow
How coding agents should operate on Hearth projects, and why.
CLI Guide
Every hearth command, permission flags, and the JSON result envelope.
MCP Guide
Connecting coding agents over MCP: setup, tool naming, permission modes.
Connect Claude Code
Point Claude Code at a Hearth project: the one-click Agent panel, manual claude mcp add / .mcp.json setup, and permission modes.
Connect Codex
Wire OpenAI’s Codex CLI to Hearth over stdio MCP: the panel launcher, codex mcp add, and the ~/.codex/config.toml format.
Connect OpenCode + Ollama
Drive Hearth entirely locally: OpenCode’s opencode.json MCP config plus a local model served by Ollama, with the tool-calling and context-window gotchas that decide whether it works.
Connect Hermes
Using Hermes with Hearth: the panel’s one-click launcher for a hermes CLI on PATH, or serving the Hermes model via Ollama/llama.cpp and driving it from an MCP-capable client like OpenCode.
Connect Any Agent
The catch-all: the canonical stdio MCP config shape, per-client config locations (Cursor, Cline, Zed, and more), and the plain CLI loop for a shell-only agent.
Reference
Project Format
Every file in a Hearth project and the schema behind it.
Architecture
Packages, the command system, and how data flows through Hearth.
Performance
The headless benchmark harness and honest before/after numbers behind the engine’s per-frame cost.
Roadmap
What is next and what is deliberately missing from the preview.
Contributing
Dev setup, ground rules, and the AI contribution policy.