OpenWAF Tasks in public beta
OpenWAF Tasks is now in public beta. A task is a scheduled job where your JavaScript (with the Playwright API) or our agent loop (with the model you choose, called through OpenWAF Egress) works in a declared OpenWAF Browsers Chrome signed with Web Bot Auth.
The browser is read-only while the task prepares. To post, submit or delete anything, the task proposes a typed list of steps; a person approves that exact list, bound by its SHA-256, and only then does the runner replay it, without your code or the model. See approvals.
The beta is free within 200 runs a month, 5 tasks and 1 run at a time per organization; browser time counts against the Browsers quota. Access is approved per organization. Tasks are in the dashboard, the API, openwaf-cloud ctl tasks and 12 new MCP tools. Start with the overview.
OpenWAF Browsers in public beta
OpenWAF Browsers, the browser sandbox for AI agents, is now in public beta. Your agent calls POST https://browsers.openwaf.dev/v1/browsers with a browser key and gets a CDP URL for Playwright, Puppeteer or the Playwright MCP server, plus a live view a person can watch or take over. Each browser is Google Chrome as shipped, in its own container in Germany, with recordings, persistent profiles and an egress proxy that logs every request and never reaches private addresses.
It is declared, not stealthy: requests are signed with Web Bot Auth by default, and there is no fingerprint spoofing, no residential proxy and no CAPTCHA solving.
The beta is free within a monthly quota of 20 browser-hours and 5 GB of egress per organization, with 2 browsers at a time per key. Access is approved per organization: every owner and admin needs a verified email and two-factor authentication. Ask from Browsers in the dashboard or write to support@openwaf.dev.
Keys, usage, sessions, the egress log and settings are in the dashboard, the API, openwaf-cloud ctl browsers and 10 new browser_* MCP tools.
This replaces the waitlist announced earlier today. See the quickstart.