# Coder Server Configuration

This section describes the configuration and features of the code-server environment.

## Accessing Code Server

- **URL:** `http://localhost:8080`
- **Authentication:** Password-based (see below for password retrieval)

### Retrieving the Code Server Password

After the first build, the code-server password is stored in:

```
configs/code-server/.config/code-server/config.yaml
```

Look for the `password:` field in that file. For example:

```yaml
password: 0c0dca951a2d12eff1665817
```

> **Note:** It is recommended **not** to change this password manually, as it is securely generated.

## Main Configuration Options

- `bind-addr`: The address and port code-server listens on (default: `127.0.0.1:8080`)
- `auth`: Authentication method (default: `password`)
- `password`: The login password (see above)
- `cert`: Whether to use HTTPS (default: `false`)

## Installed Tools and Features

The code-server environment includes:

- **Node.js 18+** and **npm**
- **Claude Code** (`@anthropic-ai/claude-code`) globally installed
- **Python 3** and tools:
  - `python3-pip`, `python3-venv`, `python3-full`, `pipx`
- **Image and PDF processing libraries**:
  - `CairoSVG`, `Pillow`, `libcairo2-dev`, `libfreetype6-dev`, `libjpeg-dev`, `libpng-dev`, `libwebp-dev`, `libtiff5-dev`, `libopenjp2-7-dev`, `liblcms2-dev`
  - `weasyprint`, `fonts-roboto`
- **Git** for version control and plugin management
- **Build tools**: `build-essential`, `pkg-config`, `python3-dev`, `zlib1g-dev`
- **MkDocs Material** and a wide range of MkDocs plugins, installed in a dedicated Python virtual environment at `/home/coder/.venv/mkdocs`
- **Convenience script**: `run-mkdocs` for running MkDocs commands easily

### Using MkDocs

The virtual environment for MkDocs is automatically added to your `PATH`. You can run MkDocs commands directly, or use the provided script. For example, to build the site, from a clean terminal we would rung:

```bash
cd mkdocs 
mkdocs build
```

## Claude Code Integration

<div class="github-widget" data-repo="anthropics/claude-code"></div>

The code-server environment comes with **Claude Code** (`@anthropic-ai/claude-code`) globally installed via npm.

### What is Claude Code?

Claude Code is an AI-powered coding assistant by Anthropic, designed to help you write, refactor, and understand code directly within your development environment.

### Usage

- Access Claude Code features through the command palette or sidebar in code-server.
- Use Claude Code to generate code, explain code snippets, or assist with documentation and refactoring tasks.
- For more information, refer to the [Claude Code documentation](https://docs.anthropic.com/claude/docs/claude-code).

> **Note:** Claude Code requires an API key or account with Anthropic for full functionality. Refer to the extension settings for configuration.

### Call Claude 

To use claude simply type claude into the terminal and follow instructions. 

```bash
claude
```

## Shell Environment

The `.bashrc` is configured to include the MkDocs virtual environment and user-local binaries in your `PATH` for convenience.

## Code Navigation and Editing Features

The code-server environment provides robust code navigation and editing features, including:

- **IntelliSense**: Smart code completions based on variable types, function definitions, and imported modules.
- **Code Navigation**: Easily navigate to definitions, references, and symbol searches within your codebase.
- **Debugging Support**: Integrated debugging support for Node.js and Python, with breakpoints, call stacks, and interactive consoles.
- **Terminal Access**: Built-in terminal access to run commands, scripts, and version control operations.

## Collaboration Features

Code-server includes features to support collaboration:

- **Live Share**: Collaborate in real-time with others, sharing your code and terminal sessions.
- **ChatGPT Integration**: AI-powered code assistance and chat-based collaboration.

## Security Considerations

When using code-server, consider the following security aspects:

- **Password Management**: The default password is securely generated. Do not share it or expose it in public repositories.
- **Network Security**: Ensure that your firewall settings allow access to the code-server port (default: 8080) only from trusted networks.
- **Data Privacy**: Be cautious when uploading sensitive data or code to the server. Use environment variables or secure vaults for sensitive information.

## Ollama Integration

<div class="github-widget" data-repo="ollama/ollama"></div>

The code-server environment includes **Ollama**, a tool for running large language models locally on your machine.

### What is Ollama?

Ollama is a lightweight, extensible framework for building and running language models locally. It provides a simple API for creating, running, and managing models, making it easy to integrate AI capabilities into your development workflow without relying on external services.

### Getting Started with Ollama

#### Staring Ollama 

For ollama to be available, you need to open a terminal and run: 

```bash
ollama serve
```

This will start the ollama server and you can then proceed to pulling a model and chatting. 

#### Pulling a Model

To get started, you'll need to pull a model. For development and testing, we recommend starting with a smaller model like Gemma 2B:

```bash
ollama pull gemma2:2b
```

For even lighter resource usage, you can use the 1B parameter version:

```bash
ollama pull gemma2:1b
```

#### Running a Model

Once you've pulled a model, you can start an interactive session:

```bash
ollama run gemma2:2b
```

#### Available Models

Popular models available through Ollama include:

- **Gemma 2** (1B, 2B, 9B, 27B): Google's efficient language models
- **Llama 3.2** (1B, 3B, 11B, 90B): Meta's latest language models
- **Qwen 2.5** (0.5B, 1.5B, 3B, 7B, 14B, 32B, 72B): Alibaba's multilingual models
- **Phi 3.5** (3.8B): Microsoft's compact language model
- **Code Llama** (7B, 13B, 34B): Specialized for code generation

### Using Ollama in Your Development Workflow

#### API Access

Ollama provides a REST API that runs on `http://localhost:11434` by default. You can integrate this into your applications:

```bash
curl http://localhost:11434/api/generate -d '{
  "model": "gemma2:2b",
  "prompt": "Write a Python function to calculate fibonacci numbers",
  "stream": false
}'
```

#### Model Management

List installed models:
```bash
ollama list
```

Remove a model:
```bash
ollama rm gemma2:2b
```

Show model information:
```bash
ollama show gemma2:2b
```

### Resource Considerations

- **1B models**: Require ~1GB RAM, suitable for basic tasks and resource-constrained environments
- **2B models**: Require ~2GB RAM, good balance of capability and resource usage
- **Larger models**: Provide better performance but require significantly more resources

### Integration with Development Tools

Ollama can be integrated with various development tools and editors through its API, enabling features like:

- Code completion and generation
- Documentation writing assistance
- Code review and explanation
- Automated testing suggestions

For more information, visit the [Ollama documentation](https://ollama.ai/docs).

For more detailed information on configuring and using code-server, refer to the official [code-server documentation](https://coder.com/docs/).