# Matrix GPT Bot

A Matrix chat bot written in Go that leverages OpenAI to provide GPT-powered conversational responses. The bot listens for messages prefixed with `gpt:` or mentions (`@gpt:<domain>`) and maintains per-user conversation history.

## Features

- **GPT Conversations**: Respond to `gpt: <your message>` commands.
- **Matrix Mentions**: Also triggers on full Matrix mention (`@gpt:domain`).
- **Contextual History**: Keeps a conversation history per user.
- **HTML Formatting**: Sends replies as plain text (no additional markup).
- **Docker Support**: Easily containerize for deployment.

## Prerequisites

- Go 1.20+ installed
- Access to a Matrix homeserver (e.g., Synapse)
- A bot user created on your Matrix server
- An OpenAI API key

## Configuration

Edit `config/config.go` to set your credentials:

```go
func Load() *Config {
    return &Config{
        Homeserver:  "https://matrix.your-server.net",
        BotUserID:   "@gpt:your-server.net",      // Your bot's user ID
        AccessToken: "YOUR_MATRIX_ACCESS_TOKEN",  // Your Matrix access token
        OpenAIKey:   "YOUR_OPENAI_API_KEY",       // Your OpenAI API key
    }
}
```

## Local Development

```bash
# Clone the repository
git clone https://git.ewnix.net/phlux/matrix-gpt-bot.git
cd matrix-gpt-bot

# Initialize modules and download dependencies
go mod init matrix-gpt-bot
go mod tidy

# Run the bot
go run .
```

Invite your bot into a Matrix room and test:

```
gpt: Hello, bot!
```

## Docker Deployment

To build a Docker image:

```bash
docker build -t matrix-gpt-bot:latest .
```

Run the container:

```bash
docker run -d   --name matrix-gpt-bot   matrix-gpt-bot:latest
```

## Kubernetes (K3s)

1. Create a namespace and pull-secret for private registries if needed.
2. Deploy with a standard Deployment manifest, mounting environment variables for credentials.

## Command Reference

- `gpt: <message>` – Sends `<message>` to OpenAI and returns the response.
- `@gpt:domain <message>` – Alternate trigger via mention.

## License

MIT © Your Name