A lightweight web application (MVP) that enhances traditional web search by using AI to generate multiple focused queries and execute them in parallel via SearxNG and filtering with AI.
See deploy/env.example
- Backend: Go 1.23, Chi router, Gorilla WebSocket, structured logging
- Frontend: Vue 3, TypeScript, Tailwind CSS, Pinia, WebSocket API
- Search: SearxNG (self-hosted)
- AI: OpenRouter API
- Deploy: Docker Compose with WebSocket-enabled Nginx
By default, SearxNG is configured to work through the searx_proxy server. If you want to disable proxy usage and make direct requests, edit the deploy/searxng_settings.yml file:
outgoing:
# Comment out or remove the proxies section
# proxies:
# all://:
# - http://searx_proxy:8380
# extra_proxy_timeout: 5.0
retries: 1MIT
