GetThatQuick

Self-hosted AI prompt workbench with speech-to-text, reusable templates, and multi-provider LLM support — all in a single Docker container.

Current docs version: v1.1.0

What is GetThatQuick?

GetThatQuick is a local-first productivity tool that lets you:

• Speak naturally and convert voice into structured prompts (local Vosk or cloud Groq/OpenAI Whisper)
• Use templates to standardize prompt formatting across sessions
• Chat with any LLM via OpenRouter, OpenAI, Ollama, GitHub Copilot, or any OpenAI-compatible endpoint
• Manage sessions with full conversation history and project grouping
• Stay fully local — no cloud dependency, no telemetry

Project Structure

GetThatQuick/
├── client/              # React 19 + Vite + Tailwind v4 frontend
│   └── src/
│       ├── api/         # API client (fetch wrappers)
│       ├── components/  # UI components (chat, sidebar, settings, etc.)
│       ├── hooks/       # React hooks (sessions, settings, templates)
│       ├── lib/         # Utilities (accent theming, class merging)
│       └── pages/       # Dashboard & Onboarding pages
├── server/              # Bun + Hono backend
│   └── src/
│       ├── routes/      # REST API endpoints
│       ├── services/    # Business logic (LLM, sessions, templates, Vosk, Copilot)
│       ├── lib/         # Constants, errors, FFI bindings, paths
│       └── ws/          # WebSocket STT handler
├── shared/              # TypeScript types & Zod schemas (shared by client + server)
├── docs-site/           # This documentation site (Docusaurus)
├── Dockerfile           # Multi-stage Docker build
├── docker-compose.yml   # Single-container deployment
├── install.sh           # One-liner installer (Linux/macOS)
└── install.ps1          # One-liner installer (Windows)

Key Features

Feature	Description
Chat Interface	ChatGPT-like UI with streaming, markdown, code blocks, thinking tokens
Local STT	Vosk — 20+ languages, runs entirely offline, no cloud APIs
Cloud STT	Groq Whisper (free, 8 hrs/day) or OpenAI Whisper — low RAM, fast setup
GitHub Copilot	OAuth device flow — access Claude, GPT-4.1, Gemini 2.5 Pro via your Copilot subscription
Templates	Create, browse, drag-to-chat. 220+ community templates from GitHub
Multi-Provider	OpenRouter, OpenAI, Ollama, LM Studio, GitHub Copilot, custom endpoints
Projects	Group sessions into projects with drag-and-drop
Plan Mode	AI asks clarifying questions before generating
Document Upload	Attach text/code files as context
Customization	Accent colors, fonts, system prompts, temperature, max tokens
Self-Hosted	Single Docker container, data at ~/getthatquick/

Tech Stack

Layer	Technology
Runtime	Bun
Server Framework	Hono
Frontend	React 19 + Vite 7 + Tailwind CSS v4
UI Primitives	Radix UI (via shadcn/ui pattern)
Icons	Lucide React
LLM Client	OpenAI SDK (compatible with any provider)
Local STT	Vosk via bun:ffi (libvosk.so)
Cloud STT	Groq Whisper / OpenAI Whisper (multipart upload)
Markdown	react-markdown + remark-gfm
Storage	Filesystem (JSON sessions, Markdown templates)
Deployment	Docker (single container, multi-stage build)

WebSocket Stack

GetThatQuick uses Bun's built-in WebSocket server.

• Upgrade path: Bun.serve(... server.upgrade(req))
• Connection type: ServerWebSocket
• Endpoint: /ws/stt
• Purpose: real-time PCM audio streaming for speech-to-text

There is no separate ws runtime dependency for the server transport layer.

License

CC BY-NC 4.0 — Free for personal and non-commercial use.