https://github.com/thinkwee/HiMe
One-Stop Personal Health AI Agent "Say Hi to Healthy Me"
https://github.com/thinkwee/HiMe
agent agentic-ai ai-agent health health-agents llm llm-agent multi-agent openclaw personal-agent personal-assistant wearable
Last synced: 22 days ago
JSON representation
One-Stop Personal Health AI Agent "Say Hi to Healthy Me"
- Host: GitHub
- URL: https://github.com/thinkwee/HiMe
- Owner: thinkwee
- License: other
- Created: 2026-04-13T18:59:10.000Z (2 months ago)
- Default Branch: main
- Last Pushed: 2026-05-05T11:58:14.000Z (about 1 month ago)
- Last Synced: 2026-05-17T07:51:16.655Z (27 days ago)
- Topics: agent, agentic-ai, ai-agent, health, health-agents, llm, llm-agent, multi-agent, openclaw, personal-agent, personal-assistant, wearable
- Language: Python
- Homepage: https://thinkwee.top/hime/
- Size: 20.4 MB
- Stars: 34
- Watchers: 0
- Forks: 5
- Open Issues: 10
-
Metadata Files:
- Readme: README.md
- Contributing: CONTRIBUTING.md
- Funding: .github/FUNDING.yml
- License: LICENSE
- Code of conduct: CODE_OF_CONDUCT.md
- Security: SECURITY.md
- Agents: AGENTS.md
Awesome Lists containing this project
- awesome-medical-ai - HiMe - square) | ⭐ C+ | Local-first personal-health AI agent with Apple Watch/iPhone ingestion, iOS/watchOS companions, Telegram/Feishu chat, reusable skills, Dockerized deployment, and tests. | (Clinical Software & EHR)
README
English | 简体中文
---
HiMe (Health Intelligence Management Engine) is a self-hosted, fully local, secure and open-source one-stop personal health AI agent platform. It understand your wearable health data in real-time and provide you with proactive insights 7/24, and of course, a cute pixel-art cat that serves as your personal health digital twin.
## Features
---
---
- Real-time wearable data ingestion from Apple Watch + iPhone, including heart rate, HRV, SpO2, sleep stages, workouts, mobility, and 50+ metrics more.
- iOS and watchOS companion apps for easy syncing health data and controlling the agent.
- Autonomous AI analysis with scheduled checks and event triggers.
- OpenClaw-style chat over Telegram, Feishu, or WeChat (via the official ClawBot plugin), with evidence-backed responses.
- Agent-generated personalised pages on demand for repeated workflows or personalised interaction. Generate your app, not learn to use it.
- Skills system for reusable analysis playbooks.
- Strong self-hosted privacy posture.
## Quick Start
Three steps. Total time: ~10 minutes.
### 1. Get IM credentials
HiMe chats with you over **Telegram**, **Feishu**, or **WeChat**. Pick one and grab the credentials before you start the server (the setup wizard will ask for them).
- **Telegram**: create a bot with [@BotFather](https://t.me/BotFather) → save the token. Send `/start` to [@userinfobot](https://t.me/userinfobot) → save your chat_id.
- **Feishu**: create a custom app at [open.feishu.cn](https://open.feishu.cn) → grab APP_ID + APP_SECRET. Invite the bot to a group → grab the open_chat_id.
- **WeChat**: no developer console needed — set `WEIXIN_GATEWAY_ENABLED=true` in `.env`, then run `python -m backend.weixin.qr_login` (or `docker exec -it hime-backend python -m backend.weixin.qr_login`) and scan the QR with WeChat → Settings → Plugins → ClawBot.
Detailed walkthrough: [`docs/INSTALL.md#im-gateway-setup`](docs/INSTALL.md#im-gateway-setup).
### 2. Start the server
```bash
git clone https://github.com/thinkwee/HiMe.git HiMe
cd HiMe
./setup.sh
```
Don't worry, just follow the quick start wizard in `setup.sh` and all the steps will be done for you. ~2-5 minutes for first build.
When it's done, the dashboard is at http://localhost:5173 — but you'll do most things through the iOS app.
### 3. Install the iOS app
- **Easy path**: install [HiMe on the App Store](https://apps.apple.com/app/id6762160735). Open Settings → Server URL → enter your host (e.g. `localhost`, your Mac's LAN IP, or `homelab.local`).
- **Build from source**: see [`docs/INSTALL.md#ios-app`](docs/INSTALL.md#ios-app).
That's it. Send the bot a message and the agent will reply.
## Update the HiMe
Pull the latest release and restart — your `.env`, `data/`, and `memory/` are preserved:
```bash
git pull
./hime.sh restart --rebuild # Docker mode: rebuilds images
./hime.sh restart --clean # Native mode: clears Python/node caches
```
If a new release adds variables, diff `.env.example` against your `.env` and fill in any missing keys. See [`docs/DEPLOYMENT.md`](docs/DEPLOYMENT.md#5-upgrades-and-backups) for backup guidance.
---
## Documentation
- [`docs/INSTALL.md`](docs/INSTALL.md) — manual setup, native dev install, public deployment, customization.
- [`docs/DEPLOYMENT.md`](docs/DEPLOYMENT.md) — LAN, public-internet, and Compose production patterns.
- [`docs/DEVELOPMENT.md`](docs/DEVELOPMENT.md) — architecture, adding tools/providers, code style.
- [`CONTRIBUTING.md`](CONTRIBUTING.md) — contribution process.
- [`SECURITY.md`](SECURITY.md) — security disclosure.
- [`PRIVACY.md`](PRIVACY.md) — privacy policy.
## Status
HiMe is research-grade software for personal use. It is not a medical device and does not provide diagnoses.
## License
HiMe is released under the [PolyForm Noncommercial License 1.0.0](LICENSE).
## Trademark
"HiMe" and the HiMe logo are trademarks of the HiMe Organisation.