browser-use makes it easier to build browser agents. Acira adds a security layer around those agents so teams can control browser actions before they happen.
Use this page when securing browser-use agents that interact with real websites, accounts, forms, files, or internal workflows.
A browser-use agent receives web page content and turns model decisions into browser actions. If a page contains malicious instructions or a workflow exceeds policy, the agent needs an enforcement layer outside the model.
Acira wraps the browser action channel with policy checks for navigation, clicks, typing, uploads, downloads, and scripts. The model can request an action, but Acira decides whether it is safe to execute.
Start by defining the domains, actions, secrets, and approval points your browser-use agent is allowed to use. Then route browser commands through Acira before they reach the browser.
Acira is designed for browser-using agents, including browser-use style automation, Playwright, Puppeteer, and raw CDP workflows.
Start with credential handling, navigation scope, destructive actions, downloads, uploads, and any action that can spend money or change data.