Browser4
An AI-native browser automation engine for autonomous agents, offering both zero-token deterministic extraction and LLM-driven extraction at large scale, controlled by a Rust CLI and MCP.
Browser4 is a browser automation engine built for autonomous agents and large-scale web automation. AI agents or humans drive real browsers through a Rust CLI, MCP, or an agentic backend, navigating, clicking, filling forms, taking snapshots, and running batch or looped jobs.
Its zero-token extraction path uses an X-SQL query language with CSS selectors for deterministic extraction from live pages or stored HTML snapshots, and a machine-learning clustering tool called WebMiner turns collections of HTML pages into spreadsheet and report views without spending any LLM tokens; a hybrid mode combines this with LLM extraction and a store of previously learned selectors and blockers. Its runtime is described as coroutine-safe and CDP-native, aimed at handling large volumes of complex page visits per day through swarm or crawl scale-out, and it also ships a "programming-agent kernel" of more than 50 tools for sandboxed shell and filesystem access that an agent can use to build or extend Browser4 itself.
It is written in Kotlin and released under the Apache-2.0 license.