开源AI工作流

Prerequisites

• Python: 3.11 or higher
• Node.js: 22 or higher
• pnpm: Package manager for frontend
• SQLite: Database for storing workspaces and research data
• Chromadb: Vector database for semantic search
• SearXNG: Self-hosted search engine (see SearXNG Setup)

Install dependencies

pip install -e .

4. **Authenticate with Hugging Face** (required for LLM model downloads)

Installation

1. Clone the repository

git clone https://github.com/introlix/introlix.git
cd introlix

1. Set up environment variables

cp .env.example .env

Edit .env and add your API keys:

```env

🐳 Running with Docker & Docker Compose (Integrated Local Run)

For running the entire application (both Next.js frontend and FastAPI backend) in a single integrated container:

1. Build the local Docker image:

   docker build -f Dockerfile.local -t introlix:local .
   

2. Run using Docker Compose: Ensure you have your .env file containing your API keys (e.g. GEMINI_API_KEY, etc.) in the root folder, then start the container:

   docker compose up -d
   

3. Access the application: - Frontend: http://localhost:8043 - Backend API Docs: http://localhost:8042/docs

To stop the container, run:

docker compose down

---

SearXNG Setup

SearXNG is a privacy-respecting metasearch engine. For Introlix to work properly, you need to configure it to return JSON results.

To install SearXNG, see Installation guide.

Modify searxng/settings.yml:

```yaml

🚀 Quick Start

• Note: If you want to use it via docker then go and see release section

Optional: Search engine

SEARCHXNG_HOST=http://localhost:8080/search

Create virtual environment

python -m venv .venv source .venv/bin/activate # On Windows: .venv\Scripts\activate

🔧 Configuration

SearXNG settings

general: instance_name: "SearXNG"

search: safe_search: 0 autocomplete: "" formats: - html - json # Important Enable JSON format

server: port: 8888 bind_address: "127.0.0.1"

**Note**: Above code is only for example. Don't replace settings.yml file. Only modify the settings.yml for enabling json.

For full template see, [searxng/blob/main/searx/settings.yml](https://github.com/introlix/searxng/blob/main/searx/settings.yml)

3. **Verify JSON output**

Test that JSON format works:

You should receive a JSON response with search results.

Important: Make sure to enable JSON format in your SearXNG settings as shown above. Introlix requires JSON responses from SearXNG to function properly.

---

Chat Interface

• Real-time conversational AI with streaming responses
• Internet search integration for up-to-date information
• Conversation history persistence
• Support for multiple LLM providers (OpenRouter, Google AI Studio)

Required: API key & JWT secret key

INTROLIX_API_KEY=123456 JWT_SECRET_KEY = my_secret

3. **Install Python dependencies**

Backend (Python/FastAPI)

• FastAPI: High-performance async web framework
• Multi-Agent System: Specialized AI agents for different tasks
• ChatAgent: Conversational interface with search
• ContextAgent: Gathers research context through questions
• PlannerAgent: Creates structured research plans
• ExplorerAgent: Searches and gathers information
• EditAgent: AI-assisted document editing
• Vector Storage: Chromadb for semantic search
• Database: SQLite for data persistence

Research Desk Workflow

The Research Desk guides you through a comprehensive research process:

1. Initial Setup: Create a research desk with your topic
2. Context Agent: AI asks clarifying questions to understand your research scope
3. Planner Agent: Generates a structured research plan with topics and keywords
4. Explorer Agent: Automatically searches the internet and gathers relevant information
5. Document Editing: AI-assisted writing and editing of your research document
6. Interactive Chat: Ask questions about your research and get AI-powered answers