> ## Documentation Index
> Fetch the complete documentation index at: https://docs.moorcheh.ai/llms.txt
> Use this file to discover all available pages before exploring further.

# moorcheh-edge up

> Pull the Docker image and start the edge server.

## Synopsis

```bash theme={null}
moorcheh-edge up [options]
```

## Options

| Flag               | Default                         | Description                                                    |
| ------------------ | ------------------------------- | -------------------------------------------------------------- |
| `--server-port`    | `8080`                          | Host port for `http://localhost:<port>`                        |
| `--server-image`   | `moorcheh/moorcheh-edge:latest` | Docker image to run                                            |
| `-y`, `--yes`      | off                             | Non-interactive: no prompts, use defaults + flags              |
| `--skip-embedding` | off                             | Skip BGE entirely (vector-only; no warm embed daemon)          |
| `--with-llm`       | off                             | Install Ollama and pull the answer model (\~400 MB)            |
| `--skip-ollama`    | off                             | Explicit search-only (same as default without `--with-llm`)    |
| `--warm-llm`       | off                             | With `--with-llm`: preload LLM in RAM (default keep-alive 24h) |

## Defaults

| Component               | Default `up`                         |
| ----------------------- | ------------------------------------ |
| Docker server           | **Yes**                              |
| BGE embed daemon (warm) | **Yes**                              |
| Ollama + LLM model      | **No** (use `--with-llm`)            |
| LLM kept in RAM         | **No** (use `--with-llm --warm-llm`) |

In an interactive terminal, `up` asks up to three questions (embedding, AI answer, keep LLM in RAM). Use `-y` or flags to skip prompts.

## Examples

```bash theme={null}
moorcheh-edge up
moorcheh-edge up -y
moorcheh-edge up --with-llm
moorcheh-edge up --with-llm --warm-llm
moorcheh-edge up --skip-embedding
moorcheh-edge up --server-port 8083
moorcheh-edge up --server-image moorcheh/moorcheh-edge:0.1.0
```

On first run with embedding enabled, `up` downloads **BGE-base-en-v1.5** (\~210 MB) and starts a warm embedding daemon (stopped by [down](/cli/runtime/down)).

With `--with-llm` on Linux, `up` also installs **Ollama** into `~/.moorcheh-edge/ollama/` and pulls **`qwen2.5:0.5b-instruct`** for [answer](/cli/answer).

## Output

On success:

```
Data directory: ~/.moorcheh-edge/data
Moorcheh Edge API: http://localhost:8080
```

With `--with-llm`:

```
[moorcheh-edge] LLM endpoint (in container): http://host.docker.internal:11434/v1
```

If the container is already running, `up` attaches and prints the same lines without starting a second container. Run `moorcheh-edge down` then `up` after upgrades so LLM settings are applied to the container.

## Requirements

* Docker running
* Pull access for `moorcheh/moorcheh-edge:latest` (or a local cached image)

## Related

* [moorcheh-edge down](/cli/runtime/down)
* [moorcheh-edge status](/cli/runtime/status)