easi.run

A browser task should not end when an agent reaches a sign-in screen. easi.run gives your AI agent a real Chrome browser on your own server, with a live view you can take over whenever a task needs you. Sign in, complete two-factor authentication, or handle an approval, then hand the browser back so the agent can continue its work.

This is useful when the work crosses a boundary that belongs to a person. An agent might need a report from an account you use, a file uploaded into a web application, or an output downloaded from a finished task. Instead of moving between separate browsers and restarting the workflow, you and the agent use the same session. After handback, the agent still needs to inspect the page and verify what happened before continuing.

Developers connect through MCP or the REST interface. The browser runs in a separate session container, and encrypted browser profiles let you reuse signed-in sessions. Live human takeover provides a practical way to handle the steps you need to do yourself. File upload and download support connects remote browser tasks with files on your own device. Recordings and an action timeline help you review the sequence of actions when a workflow needs checking.

For example, ask your connected agent to retrieve a report from an authorized dashboard. Watch the browser as it works. When a login or two-factor prompt appears, take control and complete that step yourself. Return control to the agent, let it check the current page, and have it continue to the report and download the output. The same pattern can help with document uploads and other browser tasks that mix automation with human decisions.

The main audience is developers and technical teams using coding agents such as Claude Code or Codex, or another compatible MCP client. You supply the agent, task instructions, account access, and infrastructure. easi.run supplies the browser sessions and the tools for working with them. It is self-hosted software for Linux x86-64 with Docker Engine, so it suits people comfortable running a service on their own server.

The project is open source under AGPL-3.0. There is no software usage fee; server infrastructure and model costs still apply. Read the source and installation requirements before deploying, and try the public browser interface to see how the product feels.

Source and setup: https://github.com/julianosig/easi.run Documentation: https://docs.easi.run Public browser demo: https://easi.run/demo Official product video: https://easi.run/film/en-997a5b9681d22834.mp4