Site RAG is a Chrome extension enabling users to query websites using natural language. It supports one-off queries on the current page, persistent indexing of pages or entire sites into a local vector store, and integration with local LLMs like Ollama.

How It Works

The extension operates locally within the browser, storing secrets and indexed data in browser storage. For persistent RAG, it leverages Supabase with the pgvector extension to store document embeddings and metadata. Queries are processed by generating multiple search terms ("multi-query mode") for enhanced retrieval, and conversation history is managed by including previous context in system prompts for follow-up questions.

Quick Start & Requirements

• Install: Clone the repository, run yarn install, then yarn build. Load the dist directory as an unpacked extension in chrome://extensions/.
• Prerequisites: API keys for Anthropic, OpenAI, Google GenAI, Together AI, and FireCrawl. For advanced RAG, a Supabase account and database setup with pgvector are required.
• Setup: Basic setup involves API key configuration. Advanced setup with Supabase may take 15-30 minutes.
• Links: Site RAG Demo Video

Highlighted Details

• Runs 100% locally in the browser.
• Supports integration with locally running Ollama instances.
• Utilizes Supabase with pgvector for persistent vector storage.
• Features "multi-query mode" for improved retrieval accuracy.

Maintenance & Community

The project is maintained by bracesproul. Further community or roadmap information is not detailed in the README.

Licensing & Compatibility

The README does not explicitly state a license. Compatibility for commercial use or closed-source linking is not specified.

Limitations & Caveats

The project is presented as a demo, and specific limitations regarding supported LLM providers beyond those mentioned, or the exact vector dimension compatibility for non-OpenAI embeddings, are not fully detailed. The reliance on Supabase for advanced RAG introduces an external dependency.