I'm working on HeadlessX v2.5. I know the current dashboard and a lot of the system need a real redesign — cleaner UI, better UX, and a setup that actually fits how people use this with agents. That's what I'm building now.
v2.1.2 is still what you run today. v2.5 will replace the playground-style UI and add a proper agent chat that calls operators. Live agentic browser sessions and taught recipes stay in Helium CLI — HeadlessX will not ship Interact or YAML workflows. Headed operator work (Google cookie setup, later logins) will still embed KasmVNC in the dashboard so keyboard and mouse go into the real browser window. Postgres will ship in Docker only (no Supabase). Platform install will be headlessx-setup: pick self-host or VPS, auto-detect OS (self-host on Linux/macOS/Windows; VPS Linux only), check Docker/npm/Bun, then on VPS validate DNS before start and print the Dashboard/API links. There is no developer mode.
v2.5 launches Hexium (hexium-browser / Playwright launch() → Chrome 151). Not Camoufox. Not CloakBrowser.
v2.5 will live under HeadlessXLabs. This repo stays v2.1 on @saifyxpro until that ship. Watch HeadlessXLabs and the releases here until the move.
HeadlessX is a self-hosted scraping platform with a web dashboard, protected API, queue-backed workflows, and a remote MCP endpoint.
Current live operator surfaces:
- Website operator: scrape, crawl, map, content extraction, screenshots
- Google AI Search
- Tavily
- Exa
- YouTube
- Queue jobs, logs, API keys, proxy management, and config management
- Remote MCP over
/mcp
Important operator setup notes:
- Google AI Search requires a one-time
Build Cookiesrun in the dashboard before the first search - the saved Google session is kept in the shared persistent browser profile and reused later
- the YouTube workspace is active only when
