Skip to article frontmatterSkip to article content
Site not loading correctly?

This may be due to an incorrect BASE_URL configuration. See the MyST Documentation for reference.

AI Browsers and Agentic Web Automation

Things on this page are fragmentary and immature notes/thoughts of the author. Please read with your own judgement!

This post compares AI-first web browsers and agentic browser automation tools across Linux, macOS, and Android, combining architecture analysis with real-world multi-step task benchmarks. For traditional programmable web automation frameworks (Playwright, Selenium, Puppeteer) see Web Automation, and for general browser notes see Tips on Web Browsers.

Verdict & Recommendations

Table 1:AI Browsers and Agentic Web Automation Tools Comparison

Browser / Agent Tool

Status

Linux?

Background Tabs?

Batch Scripting?

BYOK / Local LLM?

Page Perception

Primary Focus

Language

License

Popularity (Stars)

BrowserOS

Active

Y[1]

Y[2]

Y[3]

Y[4]

Accessibility snapshot ([ref=eN]) + DOM

Local MCP agent browser (browseros-neo) for complex workflows

Rust / C++[5]

AGPL-3.0

~14.3k ★

chrome-devtools-mcp

Active

Y[6]

N[7]

N[8]

Y[9]

Accessibility snapshot (uid=) + CDP

Official Chrome DevTools MCP server (powers Antigravity /browser)

TypeScript

Apache-2.0

~55.3k ★

Brave Leo

Active

Y

N

N

Y[10]

Page text & DOM context

Privacy-first daily browser with BYOM chat & summarization

C++

MPL-2.0

~24.2k ★

HARPA AI

Active

Y[11]

N

Y[12]

Y[13]

Hybrid DOM + CSS/XPath selectors

Chromium extension for web scraping, monitoring & macros

TypeScript / JS

Proprietary

N/A

Gemini in Chrome

Active

N[14]

N[15]

N

N[16]

Built-in Chrome DOM / Gemini context

Native Chrome sidebar assistant & Auto Browse

C++

Proprietary

N/A

Claude Computer Use

Active

Y[17]

N[18]

N

N[19]

Pixel screenshots + (x, y) coordinates

Full desktop GUI & OS-level visual automation

N/A

Proprietary

N/A

Perplexity Comet

Active

N[20]

Y

N

N[21]

Cloud-assisted DOM & tab context

Commercial AI-first research & assistant browser

C++

Proprietary

N/A

Deep Technical Comparison from Real-World Testing

When automating a realistic multi-step workflow—such as logging into a healthcare portal, selecting a patient, traversing a multi-screen scheduling wizard, inspecting available dates/times across providers, and stopping cleanly before final submission—four architectural differences separate BrowserOS neo from chrome-devtools-mcp (Antigravity’s built-in /browser tool), Gemini in Chrome, and Claude Computer Use.

1. Snapshot (Accessibility Tree) vs. Screenshot (Pixel Image)

A common misconception is that snapshot and screenshot are synonyms. In agentic browser tools, they are fundamentally different operations:

Why Snapshot Wins for Web Automation:

  1. Deterministic targeting: Clicking ref=e4 targets the exact DOM element regardless of window size, DPI scaling, or minor layout shifts. Coordinate clicking frequently misclicks when sticky headers, dynamic banners, or scrollbars shift the page.

  2. Speed and token efficiency: A filtered interactive snapshot (page.snapshot()) is a few hundred tokens of text generated in milliseconds; a sequence of high-resolution screenshots consumes tens of thousands of vision tokens and adds seconds of encoding/inference latency per click.

  3. No visual occlusion required: Querying the Accessibility Tree does not require the browser window to be visible on the physical monitor, whereas desktop screenshot tools capture whatever window happens to be on top.

2. Batch Scripting (run) vs. Step-by-Step LLM Round-Trips

Even though both BrowserOS neo and chrome-devtools-mcp use accessibility snapshots, their execution models differ drastically:

3. Background Execution vs. Foreground Tab Focus (bringToFront)

Can an agent work in a background tab while you keep reading or working in your active tab?

4. Connection Reliability on Linux (Flatpak Chrome & Modal Prompts)

On Linux—especially immutable distributions like Fedora Silverblue/Bluefin where Google Chrome runs inside Flatpak (com.google.Chrome)—attaching via Chrome DevTools Protocol has sharp edges:

5. BrowserOS neo vs. Gemini in Chrome & Claude Computer Use

Commercial & Alternative AI Browsers

Perplexity Comet Overview & Security Considerations

Perplexity Comet is a commercial AI-first browser built on Chromium/Blink that acts as a sidecar personal assistant:

Other Standalone & Open-Source AI Browsers

Android Support & Mobile Workarounds

BrowserOS does not support Android (and mobile support is not on its roadmap due to its desktop Chromium + local MCP server architecture). Practical ways to run BYOK/local-LLM agentic web tasks from Android include:

Table 2:Android Agentic Web Automation Workarounds

Approach

Tool

BYOK / Custom Endpoints

Autonomous Web Control

Remote Desktop Trigger

BrowserOS + OpenClaw

Full desktop BYOK / local Ollama routing

Triggers autonomous tasks on Linux/macOS via Telegram, WhatsApp, or Slack

Native Android Agent App

AIOPE (open-source) or BYOA AI Agent

Yes (OpenRouter, Ollama, OpenAI-compatible APIs)

Yes (built-in WebView automation & MCP in AIOPE)

Browser + Extension

Quetta / Mises + Nanobrowser

Yes (OpenRouter, OpenAI, Claude, Gemini, Ollama)

Yes (clicks, types, and navigates across tabs)

Local Terminal

Termux + OpenClaw

Yes (full local or BYOK API routing)

Yes (scripted & agentic browser tasks)

Standard AI Agent Skills (SKILL.md) in BrowserOS

BrowserOS natively supports the open SKILL.md standard in two complementary directions:

  1. Exposing BrowserOS as a Skill for External Coding Agents: BrowserOS ships official SKILL.md bundles—installable via npx skills add browseros-ai/browseros --skill browseros-neo—backed by its local MCP server and browseros-cli. This exposes over 25 high-level MCP tools (and 50+ browser-control actions) to external coding agents such as Antigravity, Claude Code, Codex, Gemini CLI, Cursor, and OpenCode.

  2. Running Reusable Skills Inside BrowserOS: BrowserOS natively executes reusable SKILL.md workflow definitions (shipping with built-in skills for deep research, comparisons, and form automation) and allows saving deterministic run scripts via save_skill and replaying them via run_skill.

References

Footnotes
  1. Native Linux x64 AppImage/deb packages and macOS builds.

  2. browser.open(url) and tabs (action: "new") hardcode background: true in an isolated agent tab group (<client>/<task>) with background tab throttling disabled.

  3. The run MCP tool executes multi-step async JavaScript workflows (open -> snapshot -> click -> fill -> evaluate) in a single tool call.

  4. Supports Ollama, LM Studio, OpenRouter, Anthropic Claude, Google Gemini, OpenAI, and any external agent via MCP.

  5. Companion services and CLI in Rust/TypeScript on top of a C++ Chromium fork.

  6. Works on Linux, but Flatpak Chrome requires pointing --wsEndpoint or symlinking ~/.var/app/com.google.Chrome/config/google-chrome/DevToolsActivePort, and each connection triggers Chrome’s remote-debugging confirmation modal.

  7. new_page defaults to background: false and select_page calls Page.bringToFront(), stealing focus in the active browser window.

  8. Each click, fill, or take_snapshot requires a separate MCP tool call and LLM turn.

  9. Model-agnostic MCP server; uses whatever LLM powers the host MCP client.

  10. Supports built-in models plus Bring-Your-Own-Model (BYOM) for local Ollama and custom OpenAI-compatible endpoints on desktop.

  11. Runs as a Chromium extension inside Chrome, Brave, or Edge on Linux, macOS, and Windows.

  12. Supports multi-step YAML/visual macros, scheduled monitors, and webhooks.

  13. Supports BYOK for OpenAI, Claude, Gemini, OpenRouter, and local Ollama.

  14. Officially Windows and macOS only; requires enabling experimental Glic flags on Linux.

  15. Operates as a sidebar bound to the active tab and requires user confirmation for sensitive actions.

  16. Locked to Google Gemini models; Auto Browse requires a Google AI Pro or Google AI Ultra subscription.

  17. Typically run inside a Linux Docker container with a virtual X11/Xvfb display or on macOS/Linux desktop.

  18. Controls the visible desktop screen and mouse cursor unless isolated inside a virtual display or container.

  19. Locked to Anthropic Claude models supporting the computer tool.

  20. Available on Windows, macOS, Android, and iOS; no official Linux release.

  21. Locked to Perplexity’s cloud infrastructure and account; no local LLM or BYOK support.