easi.run Web App Overview

easi.run

Developer Tools
easi.run screenshot 1

About easi.run

Learn more about this application's features and capabilities.

easi.run provides real Chrome browser sessions for external AI agents on infrastructure that you control. An external agent connects through MCP or REST and supplies the reasoning and decisions. easi.run supplies the browser environment. It is intended for developers and teams that want to give their agents access to a real browser while retaining control of the server and browser sessions.

A live viewer lets a person watch a session and take over the same browser when a workflow needs human input. This is useful for logins, two-factor authentication, approvals and manual steps. After completing the human step, the person can return control so the external agent can continue its work in that session. The human and agent work with the same browser environment rather than exchanging instructions about a separate browser.

Sessions are isolated, and saved browser profiles are encrypted. The browser environment also supports files, downloads, recordings and audio. These capabilities make it possible to include browser work and its related files in an agent workflow. The product does not provide an AI model or replace the reasoning supplied by the external agent. Access to an agent and its model remains the responsibility of the operator.

The software is available under the AGPL-3.0-only license. Running it requires your own Linux x86-64 server with systemd and rootful Docker Engine 28 or later. You provide and administer that infrastructure. Software availability does not include hosting, server capacity or AI model usage, so the operator remains responsible for those costs and for their own agent and model access.

MCP and REST are ways to connect to your installed browser stack. There is no shared public MCP endpoint for everyone to use. The stdio connection requires the installed easi.run stack; a remote connection uses the operator's own authenticated installation. This is a self-hosted browser environment for external agents rather than a hosted chatbot or an included model subscription.

The project uses Rust, Svelte, Chrome, Linux, Wayland, Docker and MCP. It is created by Juliano Gouveia. A good starting point is to read the official documentation, review the source and demo, and consider how browser sessions and human takeover would fit your own agent setup. No independent runtime or client compatibility test is claimed by this directory description. This description was prepared with AI assistance.

Ratings & Reviews

No reviews yet. Be the first to rate!

self-hostedmcpbrowser-automationopen-source