I got tired of gluing together proxy rotation, fingerprint spoofing, job queues, and deploy scripts every time I needed to run browser automation (bots/scrapers) at scale. So I built an orchestration layer for it.
You write a Python script (Playwright, Selenium, or nodriver), add a JSON config, and Armada distributes the work across K8s pods. Jobs go through RabbitMQ, per-pod configs are seeded via Redis, proxies and fingerprints are handled by dedicated services. Scaling from 1 to 100+ workers needs no code changes.
It also ships with Fantomas, a browser automation lib on top of nodriver, physics-based mouse curves (WindMouse), randomized keystrokes, and an XDO mode that fires OS-level events via xdotool.
You write a Python script (Playwright, Selenium, or nodriver), add a JSON config, and Armada distributes the work across K8s pods. Jobs go through RabbitMQ, per-pod configs are seeded via Redis, proxies and fingerprints are handled by dedicated services. Scaling from 1 to 100+ workers needs no code changes.
It also ships with Fantomas, a browser automation lib on top of nodriver, physics-based mouse curves (WindMouse), randomized keystrokes, and an XDO mode that fires OS-level events via xdotool.
Code: https://github.com/armadasvc/armada (AGPL v3)
Docs: https://armada.services/docs
Reach out anytime.