Lighthouse for the AI Age - A free, open-source website auditing tool that evaluates and optimizes websites for AI agent interactions.
With 1 billion+ monthly AI crawler requests (GPTBot, ClaudeBot, PerplexityBot), optimizing for AI visibility is no longer optional. SoberAI is the first specialized tool for ensuring your website works perfectly with AI agents, LLMs, and autonomous browsing systems.
| Category | Weight | Description |
|---|---|---|
| SSR Readiness | 20% | Server-side rendering for AI agents that don't execute JavaScript |
| Schema Coverage | 20% | Structured data (Schema.org) for AI understanding |
| Semantic Structure | 20% | HTML5 semantic elements and heading hierarchy |
| Content Extractability | 20% | Text readability and accessibility for LLMs |
| Machine Readability | 20% | AI crawler access (robots.txt, llms.txt, sitemap, OpenGraph) |
AI-powered recommendations are generated using your choice of LLM provider (Ollama or OpenAI).
Download the latest release for your platform from the Releases page.
# Clone and install
git clone https://github.com/nitishagar/sober-ai.git
cd sober-ai
npm install
# Build frontend
cd frontend && npm install && npm run build && cd ..
# Set up database
npx prisma generate --schema=src/db/schema.prisma
npx prisma migrate deploy --schema=src/db/schema.prisma
# Start the server
node src/api/server.js
# Open http://localhost:3000npm run electron:devFor AI-powered recommendations, install Ollama and pull a model:
ollama pull qwen3:4bOr configure OpenAI in Settings with your API key.
- Navigate to the Audit page
- Enter a URL (e.g.,
https://vercel.com) - Click Run Audit and watch real-time progress (auto-reconnects on network interruption)
- Review scores across 5 categories, findings, and AI recommendations
- Compare reports over time to track improvements
- Configure your LLM provider in Settings
Electron Shell
├── React Frontend (BrowserWindow)
├── Express.js Backend (in-process)
├── SQLite Database (Prisma ORM)
├── Playwright (browser automation)
└── LLM Provider (Ollama local/cloud or OpenAI)
- Desktop: Electron
- Backend: Express.js + Node.js 22+
- Frontend: React + Vite
- Database: SQLite via Prisma
- Browser: Playwright
- LLM: Ollama (local/cloud) or OpenAI
npm test # Run all tests (185 tests, 18 suites)
npm run test:unit # Unit tests only
npm run test:integration # Integration tests only
npm run test:ci # CI mode (all tests, force exit)
npm run verify # Run tests + lintThe test suite covers:
- Unit tests: All 5 audit classes (score boundaries, severity, findings), scorer, validator, LLM analyzer, provider factory
- Integration tests: SSE streaming (reconnect, error paths), audit pipeline, settings, reports (CRUD + compare), provider routing, health
Full documentation available at nitishagar.github.io/sober-ai.
We welcome contributions! See the Contributing Guide for development setup and guidelines.
MIT License - see LICENSE for details.
Copyright 2025-2026 Nitish Agarwal
Version 0.3.0 | Changelog