mirror of
https://github.com/wu736139669/hapi.git
synced 2026-08-05 06:24:37 +00:00
weishu
958e78dc6e
The project is no longer primarily a bot - it now provides HTTP API, Socket.IO, SSE, and web app backend services. This rename reflects the broader scope of server functionality. Updates references in: - cli/src/configuration.ts (core config) - cli/src/ui/logger.ts (remote logging) - cli/src/ui/doctor.ts (diagnostics) - cli/src/commands/auth.ts (auth status) - cli/src/daemon/daemon.integration.test.ts (test comments) - README.md (quickstart docs) - cli/README.md (CLI docs) - cli/CLAUDE.md (dev docs)
105 lines
2.5 KiB
Markdown
105 lines
2.5 KiB
Markdown
# hapi
|
|
|
|
HAPI means "哈皮," a Chinese transliteration of [happy](https://github.com/slopus/happy), great credit to the original Happy project.
|
|
|
|
Run Claude Code / Codex / Gemini sessions locally and control them remotely through a Web / PWA / Telegram Mini App.
|
|
|
|
## Features
|
|
|
|
- Start AI coding sessions from any machine.
|
|
- Monitor and control sessions from your phone or browser.
|
|
- Approve or deny tool permissions remotely.
|
|
- Browse files and view git diffs.
|
|
- Track session progress with todo lists.
|
|
- Supports multiple AI backends: Claude Code, Codex, and Gemini.
|
|
|
|
## Quickstart
|
|
|
|
1. Start the server on a machine you control:
|
|
|
|
```bash
|
|
npx @twsxtd/hapi server
|
|
```
|
|
|
|
> Alternatively, download the prebuilt binary from [Releases](https://github.com/tiann/hapi/releases) and run `hapi server`.
|
|
|
|
2. If the server has no public IP, expose it over HTTPS:
|
|
- Cloudflare Tunnel: https://developers.cloudflare.com/cloudflare-one/connections/connect-networks/
|
|
- Tailscale: https://tailscale.com/kb/
|
|
|
|
3. Run the CLI on the machine where you want sessions:
|
|
|
|
```bash
|
|
# If the server is not on localhost:3006
|
|
export HAPI_SERVER_URL="https://your-domain.example"
|
|
|
|
npx @twsxtd/hapi
|
|
```
|
|
|
|
4. Open the UI in a browser at the server URL and log in with `CLI_API_TOKEN`.
|
|
|
|
## Telegram Mini App (optional)
|
|
|
|
To use Telegram for notifications and the Mini App:
|
|
|
|
1. Create a bot with @BotFather and get the token.
|
|
|
|
2. Expose your server over HTTPS (Cloudflare Tunnel, Tailscale, etc.).
|
|
|
|
3. Add environment variables:
|
|
|
|
```
|
|
WEBAPP_URL="https://your-domain.example"
|
|
TELEGRAM_BOT_TOKEN="..."
|
|
```
|
|
|
|
4. Start the server and send `/start` to the bot to get your chat ID.
|
|
|
|
5. Add your chat ID and restart:
|
|
|
|
```
|
|
ALLOWED_CHAT_IDS="12345678"
|
|
```
|
|
|
|
6. Run `/app` in the bot chat to open the Mini App.
|
|
|
|
## Multi-agent support
|
|
|
|
- `npx @twsxtd/hapi` - Start a Claude Code session.
|
|
- `npx @twsxtd/hapi codex` - Start an OpenAI Codex session.
|
|
- `npx @twsxtd/hapi gemini` - Start a Google Gemini session.
|
|
|
|
## CLI config file
|
|
|
|
You can store the token in `~/.hapi/settings.json` instead of an env var.
|
|
Environment variables take priority over the file.
|
|
|
|
## Requirements
|
|
|
|
- Claude CLI installed and logged in (`claude` on PATH) for Claude Code sessions.
|
|
- Bun if building from source.
|
|
|
|
## Build from source
|
|
|
|
```bash
|
|
bun install
|
|
bun run build
|
|
```
|
|
|
|
Build a single binary with embedded web assets:
|
|
|
|
```bash
|
|
bun run build:single-exe
|
|
```
|
|
|
|
## Docs
|
|
|
|
- `cli/README.md` - CLI usage and config
|
|
- `server/README.md` - Server setup and architecture
|
|
- `web/README.md` - Web app behavior and dev workflow
|
|
|
|
## License
|
|
|
|
- cli: MIT
|
|
- others: LGPLv2
|