# obm 🚢

**Deploy your own AI agent to the cloud in about five minutes.**

Just answer a few questions and your agent is live in the cloud.

---

## What is this?

`obm` is a command-line tool that walks you through setting up an AI agent on cloud infrastructure. It asks you things like "which cloud provider?" and "which AI model?" — then validates your answers on the spot, writes the config, and hands it off to Terraform to build everything.

It supports two agent frameworks:

- **Hermes Agent** — "Self-improving AI agent" built by Nous Research. Built-in learning loop, creates skills from experience, improves them during use, nudges itself to persist knowledge, and builds a deepening model of who you are across sessions.
- **OpenClaw** — "The AI that actually does things." Clears your inbox, sends emails, manages your calendar, checks you in for flights. Action oriented.

Both run on either **Hetzner Cloud** or **DigitalOcean**.

---

## Quick start

### Install

**Mac or Linux (one command):**

```bash
curl -fsSL https://raw.githubusercontent.com/openboatmobile/obm/main/scripts/install.sh | sh
```

**Or download a binary** from the [releases page](https://github.com/openboatmobile/obm/releases/latest) and put it somewhere on your PATH.

**Or build from source** (requires Go 1.22+):

```bash
git clone https://github.com/openboatmobile/obm.git
cd obm
make build
```

### Deploy your agent

```bash
obm deploy
```

That's it. You'll get an interactive walkthrough that looks like this:

```
🚢 OpenBoatmobile — Deploy your AI agent

Step 1: Agent Framework
  [1] Hermes Agent (Nous Research) — Python-based, highly configurable
  [2] OpenClaw — Node.js-based, simpler setup

Step 2: Cloud Provider
  [1] Hetzner Cloud — from €4.49/mo (recommended, ~70% cheaper)
  [2] DigitalOcean — from $6/mo (wider region availability)

Step 3: Provider API Token
  Get yours at: https://console.hetzner.cloud/ → Security → API Tokens
  Token: ********
  ✓ Token validated

...
```

At the end, you'll see a summary with cost estimate and get asked to confirm. Say yes, and `obm` writes your config and kicks off Terraform.

### Other commands

| Command | What it does |
|---------|-------------|
| `obm deploy` | Interactive walkthrough to set up a new agent |
| `obm validate` | Checks your existing config and API keys |
| `obm status` | Shows the state of your current deployment |
| `obm destroy` | Tears down your infrastructure (asks first, don't worry) |
| `obm version` | Prints the version |

### Non-interactive mode (for automation)

If you're running this in CI/CD or just don't want the prompts:

```bash
obm deploy --config deploy.yaml
```

See [`deploy.yaml.example`](deploy.yaml.example) for the full config file format.

---

## What you'll need

Before running `obm deploy`, have these ready:

1. **A cloud provider account** — [Hetzner Cloud](https://console.hetzner.cloud/) or [DigitalOcean](https://cloud.digitalocean.com/). Hetzner is cheaper; DigitalOcean has more data center locations.
2. **An API token** from your cloud provider. You can generate one in their dashboard.
3. **An AI model API key** — Venice AI, OpenRouter, OpenAI, or Anthropic. This is the "brain" your agent will use.
4. **An SSH public key** uploaded to your cloud provider (so you can log into your server later).

Optional but recommended:
- **Tailscale** account for VPN access to your server
- **Discord bot token** if you want your agent to chat on Discord

---

## How much does it cost?

The server cost depends on your cloud provider and server size. `obm` shows you the estimated monthly cost before you commit.

**Rough starting points:**

| Provider | Smallest option | Good for |
|----------|----------------|----------|
| Hetzner | €4.49/mo (2 vCPU, 4 GB RAM) | Most agents |
| DigitalOcean | $6/mo (1 vCPU, 1 GB RAM) | Light use |

The AI model API costs are separate and depend on your usage.

---

## What happens under the hood?

`obm` generates a `.env` file that [Terraform](https://terraform.io) reads to provision your server, install the agent software, and configure everything. You don't need to know Terraform — `obm` handles it.

The Terraform configs live in the [openboatmobile-ai](https://github.com/openboatmobile/openboatmobile-ai) repo. `obm` is the friendly CLI wrapper around them.

---

## Project status

`obm` is actively developed and functional for the core deploy workflow. Here's what works and what's coming:

**Working now:**
- Interactive deploy walkthrough (8 steps)
- API key validation for cloud providers and inference providers
- SSH key listing from Hetzner
- `.env` file generation
- Config validation (`obm validate`)
- Infrastructure teardown (`obm destroy`)
- Non-interactive mode with YAML config
- Cross-compiled binaries (Linux, macOS, Windows)
- `curl | sh` installer

**Coming soon:**
- `obm status` — SSH health checks on your deployment
- DigitalOcean provider validation
- Cost estimation display
- Resumable deploy (pick up where you left off)

---

## For developers

Building, testing, and contributing — see [DETAILS.md](DETAILS.md) for the full technical reference and [CONTRIBUTING.md](CONTRIBUTING.md) for the contribution guide.

## For AI agents

If you're an AI agent reading this to learn about the project, check out [LLMs.md](LLMs.md) — it's written specifically for you.

---

## License

MIT