EMQX Knowledge Base

A knowledge base application for EMQX that allows users to ask questions and get answers based on the knowledge base. It also includes a log analysis feature for EMQX logs.

Features

• Ask questions and get answers based on the knowledge base
• Upload files to provide additional context for questions
• Analyze EMQX logs to identify issues and get troubleshooting recommendations
• Slack integration for asking questions and getting answers

Technologies

• Backend: FastAPI, SQLAlchemy, LlamaIndex
• Frontend: Vue.js, Tailwind CSS
• Database: PostgreSQL
• LLM Support: OpenAI, Anthropic, Cohere, Gemini, HuggingFace (via LlamaIndex)

Setup

1. Clone the repository
2. Install uv (if not already installed):

   curl -LsSf https://astral.sh/uv/install.sh | sh
3. Set up environment variables:

   cp .env.example .env
4. Edit the .env file to set your LLM API key and other configuration options
5. Run the application:

   uv run main.py

Environment Variables

LLM Configuration

• LLM_PROVIDER: The LLM provider to use (default: openai, options: openai, anthropic, cohere, gemini, huggingface)
• LLM_API_KEY: Your LLM API key
• LLM_MODEL: The model to use (depends on the provider, default: gpt-4o)
• LLM_TEMPERATURE: The temperature to use for generation (default: 0.7)
• EMBEDDING_MODEL: The embedding model to use (default: text-embedding-3-small)
• EMBEDDING_DIMENSION: The dimension of the embeddings (default: 1536)

Database

• DATABASE_URL: The database URL (default: postgresql://postgres:postgres@localhost:5432/knowledge_base)

Slack Integration

• SLACK_BOT_TOKEN: Your Slack bot token (optional)
• SLACK_APP_TOKEN: Your Slack app token (optional)
• SLACK_SIGNING_SECRET: Your Slack signing secret (optional)
• SLACK_TEAM_ID: Your Slack team ID (optional)

Feature Flags

• ENABLE_SLACK: Whether to enable Slack integration (default: false)

Logging and Debugging

• LOG_LEVEL: The logging level to use (default: INFO, options: DEBUG, INFO, WARNING, ERROR, CRITICAL)
• LLAMA_INDEX_VERBOSE: Whether to show verbose LlamaIndex logs, including workflow step execution (default: false)

API Endpoints

• GET /api/health: Health check endpoint

Web UI

The web UI is available at http://localhost:8000 and includes a unified chat interface that can:

• Answer questions about EMQX Knowledge Base
• Analyze EMQX logs for troubleshooting
• Process file uploads for log analysis
• Handle EMQX API credentials to query broker information

Development

To set up the development environment:

1. Install frontend dependencies:

   cd web
   npm install
2. Run the backend:

   uv run main.py
3. Run the frontend:

   cd web
   npm run dev

License

This project is licensed under the MIT License - see the LICENSE file for details.