sleepy
f0ce25991f
- Add remark-gfm for GitHub Flavored Markdown support - Create dedicated MarkdownContent component - Improved message bubble padding (16px 20px) for better spacing - Code blocks now in separate box with dark background (#1e1e1e) - Code blocks show language label in blue accent header - Fixed list indentation to align properly with message text - Lists have consistent spacing (margin/padding) - Better blockquote styling with background and border - Improved table styling with borders and alternating row colors - Support for all common Markdown elements (headings, links, bold, italic, hr) - All 52 tests passing
Light Chat
A lightweight Android chat application built with React and Capacitor for connecting to locally or remotely hosted LLMs.
Features
- LLM API Integration: Connect to any OpenAI-compatible API endpoint
- Session Management: Multiple chat sessions with add/remove functionality
- Dark Theme: Sleek black/dark UI design
- Conversation Export: Save chats as text, log, or markdown files
- Custom System Prompts: Inject your own system prompts for personalized interactions
- Model Configuration: Custom model name support for local LLMs
- Offline Ready: Capacitor wrapper for native Android APK generation
Tech Stack
- Frontend: React + TypeScript
- Mobile Wrapper: Capacitor
- Build Tool: Vite
- Package Manager: pnpm
- Testing: Vitest + React Testing Library
- Android Studio: For final APK generation
Getting Started
Prerequisites
- Node.js (v18+)
- pnpm
- Android Studio (for APK building)
- Capacitor CLI
Installation
# Install dependencies with pnpm
pnpm install
# Start development server
pnpm dev
Build for Android
# Type check TypeScript code
pnpm type-check
# Build the web app
pnpm build
# Sync with Capacitor
npx cap sync
# Open in Android Studio
npx cap open android
Project Structure
The codebase follows these principles:
- Simple: Keep things straightforward
- Modular: Separate concerns into distinct modules
- Reusable: Abstract common functionality
- Tested: Unit tests for all components
Contributing
When adding features:
- Look for existing similar functions to abstract
- Write unit tests
- Keep components small and focused
- Follow the modular architecture