LLM Browserbrowser automation bridge

LLM Browser

A Chrome-like desktop browser built to be driven by humans and LLMs. Browse normally, then let an AI agent navigate, click, fill forms, inspect pages, take screenshots, and report back through a visible bridge panel with vertically resizable prompt and chat areas.

Subscribe for software access See every screenshot

Electron desktop app HTTP + WebSocket API Port 9224 control server Bring-your-own LLM

LLM Browser settings panel with API key provider routing, model selection, reasoning controls, and browser settings. — **Connection settings**Provider routing, model controls, system prompt, browser options, and save/test actions stay visible inside the browser.

A local browser control surface for AI agents.

LLM Browser exposes browser actions as JSON-RPC tools while preserving a normal browser interface for the person supervising the work.

9224Local control server port for HTTP and WebSocket clients

6LLM connection modes for hosted, local, CLI, OAuth, and terminal routes

20+Browser tools for tabs, navigation, content, forms, screenshots, and cookies

LLM-user bridge

The AI works in the browser you can see.

The bridge panel keeps conversation, tool output, and browser state in the same window, so automation is observable instead of hidden in a remote session.

Human browsing stays intact

Tabs, address bar, bookmarks, history, downloads, settings, and dark/light themes behave like a familiar desktop browser.

Agents get tool access

LLMs can navigate, click selectors, fill fields, execute JavaScript, read page content, scroll, manage cookies, and switch tabs.

Supervision remains visible

The chat bridge shows prompts and assistant responses while the browser itself stays front and center, with vertical resizing for both the prompt box and LLM-user chat box.

Connection modes

Use the model route that fits the task.

LLM Browser can connect to cloud APIs, local servers, CLI commands, OAuth providers, Terminal templates, or local AI workflows.

API Key

OpenAI-compatible providers with model, reasoning, token, endpoint, and system-prompt controls.

Local Server

Point the browser at local OpenAI-compatible endpoints such as Ollama, LM Studio, llama.cpp, or vLLM.

CLI Pipe and Terminal

Pipe prompts through local command-line or terminal templates when the toolchain already lives on your machine.

OAuth and Local AI

Route through OAuth providers or local AI setup paths when credentials or model files need a different flow.

LLM Browser OAuth settings tab with provider, client ID, endpoint, auth URL, token URL, scope, and reasoning controls. — **OAuth route**Connection tabs make provider-specific setup visible and editable without leaving the browser.

Latest screenshots

Every current LLM Browser screenshot.

The full media set shows the browser shell, settings routes, compact bridge states, and real browser-agent sessions.

LLM Browser open on Google with the bridge panel and prompt composer at the bottom. — **Browser home**Chrome-like browsing with the LM_Browser bridge docked below.

LLM Browser API key connection settings. — **API key settings**Hosted provider routing with model and reasoning controls.

LLM Browser local server connection settings. — **Local server settings**Connect to local OpenAI-compatible endpoints.

LLM Browser CLI pipe settings. — **CLI pipe settings**Run prompts through command-line templates.

LLM Browser OAuth settings. — **OAuth settings**Configure provider OAuth metadata and token flow.

LLM Browser local AI settings. — **Local AI settings**Local model setup inside the same browser settings surface.

LLM Browser showing a browser-agent search task with bridge messages. — **Browser search task**The bridge conversation stays attached while the page changes.

LLM Browser controlling Gmail with a visible assistant bridge. — **Inbox agent session**A live web app session with bridge status and prompts.

LLM Browser working in Gmail spam search with the LLM bridge below. — **Spam cleanup session**The prompt box and conversation remain visible while the browser works.

Compact LLM Browser bridge panel. — **Resizable bridge**The LLM-user chat area can be adjusted vertically to preserve browser space.

Compact LLM Browser chat composer area. — **Resizable prompt box**The user prompt composer can also be adjusted vertically for longer instructions.

Software subscription

Subscribe for access to LLM Browser and every listed MultiAgentAI product.

LLM Browser is part of the MultiAgentAI software subscription. Subscribers receive download links by email as products become ready.

Subscribe $10/month See all software

LLM Browser$10/mo

Your subscription covers LLM Browser plus every listed software product as it launches.