https://github.com/ayfri/deep-research
My own implementation of DeepResearch.
https://github.com/ayfri/deep-research
ai-chat deep-research openai perplexity
Last synced: 8 months ago
JSON representation
My own implementation of DeepResearch.
- Host: GitHub
- URL: https://github.com/ayfri/deep-research
- Owner: Ayfri
- License: gpl-3.0
- Created: 2025-02-24T16:01:04.000Z (over 1 year ago)
- Default Branch: main
- Last Pushed: 2025-04-24T09:31:55.000Z (about 1 year ago)
- Last Synced: 2025-04-24T10:24:38.664Z (about 1 year ago)
- Topics: ai-chat, deep-research, openai, perplexity
- Language: Svelte
- Homepage:
- Size: 931 KB
- Stars: 8
- Watchers: 1
- Forks: 1
- Open Issues: 0
-
Metadata Files:
- Readme: README.md
- License: LICENSE
Awesome Lists containing this project
README
# Perplexity Fast Chat 🚀
**A modern web interface to interact with Perplexity AI and OpenAI models, featuring a unique deep research mode.**
[Features](#features-) • [Preview](#preview-) • [Installation](#installation-%EF%B8%8F) • [Usage](#usage-) • [Technologies](#technologies-used-) • [Project Structure](#project-structure-) • [Contribution](#contribution-)
## Table of Contents 📑
- [Features](#features-)
- [Preview](#preview-)
- [Installation](#installation-%EF%B8%8F)
- [Usage](#usage-)
- [Technologies Used](#technologies-used-)
- [Project Structure](#project-structure-)
- [Deep Research Mode](#deep-research-mode-)
- [Research Settings](#research-settings-%EF%B8%8F)
- [Contribution](#contribution-)
- [License](#license-)
## Features ✨
### Models Support
- 🤖 **Perplexity AI Models**
- Sonar Reasoning Pro
- Sonar Reasoning
- Sonar Pro
- Sonar
- R1-1776
- 🧠 **OpenAI Models for Deep Research**
- o3-mini (high reasoning)
- o1-mini
- o1-preview
- o1
- 4o
### Core Features
- 🔍 **Exclusive Deep Research Mode**
- Automatic breakdown of complex questions
- Multi-step in-depth research
- Final synthesis of results
- Configurable question count (auto or manual)
- Multiple research phases
- 💬 **Modern Chat Interface**
- Conversation history
- Markdown support
- Citations and references
- "Thinking" mode with animations
- 🎨 **Sleek Design**
- Responsive interface
- Dark mode
- Smooth animations
- Support for links and references
- ⚙️ **Flexible API Key Management**
- Configure API keys directly in the UI (stored in `localStorage`)
- Fallback to server-side environment variables if UI keys are not set
## Preview 📸
### Deep Research Mode
### Settings Modal
### Chat Interface
## Installation 🛠️
### Prerequisites
- Node.js (v16+)
- pnpm (v6+)
- Perplexity API key (Optional, can be set in UI)
- OpenAI API key (Optional, can be set in UI)
### Setup Steps
1. **Clone the repository**
```bash
git clone https://github.com/your-username/perplexity-fast-chat.git
cd perplexity-fast-chat
```
2. **Install dependencies**
```bash
pnpm install
```
3. **Configure environment variables**
Create a `.env` file at the project root:
```env
PERPLEXITY_API_KEY=your_perplexity_api_key
OPENAI_API_KEY=your_openai_api_key
```
*Note: These server-side keys are optional if you plan to configure them via the UI.*
4. **Start the development server**
```bash
pnpm dev
```
Your app should now be running at [http://localhost:5173](http://localhost:5173)
## Usage 🔮
1. **Select a model** from the dropdown menu
2. **Choose a mode**:
- Regular chat for quick answers
- Deep Research for complex questions
3. **Type your question** or prompt
4. For Deep Research mode, configure:
- OpenAI model for analysis
- Question count (auto or manual)
5. **(Optional) Configure API Keys**: Click the "API Key Settings" button in the bottom-left navbar to enter your Perplexity and OpenAI keys. These will be saved in your browser.
## Technologies Used 🧰
[](https://kit.svelte.dev/)
[](https://tailwindcss.com/)
[](https://www.typescriptlang.org/)
- **[SvelteKit](https://kit.svelte.dev/)** - Web framework
- **[TailwindCSS](https://tailwindcss.com/)** - CSS framework
- **[Perplexity AI API](https://docs.perplexity.ai/)** - Chat API
- **[OpenAI API](https://platform.openai.com/)** - API for Deep Research mode
- **[Marked](https://marked.js.org/)** - Markdown parsing
- **[Lucide Icons](https://lucide.dev/)** - Icon set
## Project Structure 📁
```
/
├── src/
│ ├── routes/ # Application routes
│ │ ├── +page.svelte # Main page
│ │ └── api/ # API endpoints
│ │ ├── chat/ # Standard chat endpoint
│ │ ├── deep-research/ # Deep Research endpoint
│ │ └── name/ # Name generation endpoint
│ └── lib/ # Components and utilities
│ ├── components/ # Svelte components
│ │ └── common/ # Common small components
│ ├── helpers/ # Helper functions
│ ├── stores/ # Svelte stores
│ └── types/ # TypeScript types
├── images/ # Screenshots and images
├── .env # Environment variables
└── package.json # Dependencies and scripts
```
## Deep Research Mode 🔬
The Deep Research mode is a unique feature that:
1. **Analyzes** your question using OpenAI models (configurable)
2. **Breaks down** into relevant sub-questions (auto or manual count)
3. **Researches** each aspect using Perplexity AI
4. **Organizes** research into multiple phases when needed
5. **Synthesizes** the results into a comprehensive answer
6. **Provides** detailed timing and progress information
## Research Settings ⚙️
The Deep Research mode includes configurable settings:
| Setting | Options | Description |
|---------|---------|-------------|
| OpenAI Model | o3-mini, o1-mini, o1-preview, o1, 4o | Select model for question analysis |
| Question Count | Auto / Manual (3-40) | Control how many sub-questions are generated |
| Reasoning | High / Standard | Different reasoning approaches based on model |
## Contribution 🤝
### API Key Management
This application supports two ways to provide API keys:
1. **UI Configuration (Recommended for Local Use)**: Use the "API Key Settings" modal (accessible from the navbar) to enter your Perplexity and OpenAI keys. These are stored securely in your browser's `localStorage` and sent with each request.
2. **Server Environment Variables**: Set `PERPLEXITY_API_KEY` and `OPENAI_API_KEY` in a `.env` file at the project root. This is suitable for deployed environments or if you prefer not to store keys in the browser.
The application prioritizes keys set in the UI. If UI keys are not found, it falls back to using the environment variables.
Contributions are welcome! Please follow these steps:
1. **Fork** the project
2. **Create** a branch (`git checkout -b feature/improvement`)
3. **Commit** your changes (`git commit -m 'Add feature'`)
4. **Push** to the branch (`git push origin feature/improvement`)
5. **Open** a Pull Request
Please ensure your code follows the project's coding style and include appropriate tests.
## License 📄
This project is licensed under the MIT License - see the [LICENSE](LICENSE) file for details.
---
Made with ❤️ for AI research enthusiasts