Best headless browser for AI agents. Lite, Fast, High-Compatibility. Built in Rust
Browser agents and automation for LLMs
Many tasks still have no API: filling in a form, checking an order status, pulling data from an internal dashboard. Browser agents let a language model drive a real browser, reading the page and deciding where to click and what to type.
Some projects are libraries you call from code, others are complete services or sandboxes that control a whole computer. Reliability on long tasks is the main difference, so check how each one recovers from a failed step.
Updated 2 min ago
- Repos tracked
- 65
- New this week
- +13
- added to the list
- Stars gained in 30 days
- +36.2k
- across 43 ranked repos
- Updated
- Oct 20, 2026
Rising newcomers
22 newcomersCreated in the last 90 days and gaining at least 20 stars this week.
onehumanbeing/esp32-iphone-agentOpen onehumanbeing/esp32-iphone-agent on GitHub (opens in new tab)
NewLet an AI model use a real iPhone with a $7 ESP32: AirPlay eyes, USB HID hands, any OpenAI-compatible vision model. No Mac, no developer account, no jailbreak.
Scout: the open-source browser agent, by Outcrawl. Mirrored automatically from Outcrawl's own repository.
Codex computer use, decoupled from the app, for usage inside any harness.
nanoMuse: an open-source personal agent for every device you own. One agent with a name and a look of its own that does things, keeps working while the app is closed, remembers you, and asks before anything you cannot undo. On Android, iPhone and iPad, Windows / macOS / Linux and in the browser; your phone's screen and your computer as its hands.
Fastest and cheapest web agent
Open-source Grok Bot alternative with a virtual machine that bots can use
An open-source stack for generalist computer-use agents, combining code execution, GUI interaction, and cross-device control.
Open-source AI coworkers that each get a computer of their own: a browser, files and tools, with every action decided before it happens and recorded after. Bring any AG-UI agent.
源码级 Chromium 指纹内核,检测站按普通 Chrome 评分。Source-level Chromium fingerprint kernel that passes bot detection tests. Not JS injection — C++ patches compiled into the binary.
Nosis is a lightweight personal agent runtime: per-session execution boundaries with OS-level sandboxing, computer use, a full built-in toolset, Skill / Plugin / MCP extensions, one runtime shared by GUI and TUI, and file-persisted sessions, plans, and memory.
Accessibility-first browser & computer use for AI agents — one MCP server for every harness (Codex, Claude Code, OpenCode, Antigravity). Instruments the DOM/accessibility tree via CDP instead of screenshot loops: diffed snapshots, stable element refs, trusted input, typed errors. Zero dependencies
touri: open-source desktop app for AI agent bots (Codex, Claude, OpenCode) with isolated VMs, browser and computer use.
A floating chat bubble that lets Claude Code, Codex or OpenCode use your Android phone. Runs inside the app on your own subscription or free models. No adb, no PC, no root.
Chrome sidebar extension that lets DeepSeek Harness operate your browser directly, no vision capabilities required. 一款 Chrome 侧边栏扩展程序,可让 DeepSeek Harness 直接操控您的浏览器,无需视觉能力。
The simplest, thinnest harness that gives an LLM complete freedom to control a Mac.
feder-cr/invisible_playwright_mcpOpen feder-cr/invisible_playwright_mcp on GitHub (opens in new tab)
NewPlaywright MCP server undetected by anti-bots and captchas: AI agent browses the web on anti-detect stealth Firefox, Python, undetected browser automation, scraping, computer use.
Jev-powered model routing, memory, compaction, skill selection, computer and browser use for Hermes agents (also Claude Code and Codex)
The long-horizon computer-use harness. Run AI agents across desktop apps and the CLI for extended periods while preserving task state and making reliable progress on complex workflows. Features fresh-context execution, durable verified state, independent auditing, recoverable progress, and native Claude Code / Codex / OpenClaw integration.
Bops: a team of AI bots that run your business ops, each with its own computer, email and phone number. bops.bot
Computer use for about $0.0002 a step: OCR the screen, classify the next action with TypeSafe, click. macOS.
Let your AI agents use your Chrome, signed in as you, in background tabs. By Alexey Indeev · blog: read.aindeev.com · X: @AlexeyIndeev
How this list is ranked
60% momentum, 40% health. Newcomers: created in the last 90 days with ≥ 20 stars this week.
How the score is computed
- Health (0–100) reflects release cadence, recent commit activity and the number of active contributors.
- Momentum (0–100) is stars gained in 30 days ÷ √(stars before), scaled within this collection, so a small project growing fast can outrank a large one.
- Newcomers are listed only in their own block, so a repo never appears twice.
- Several repos from the same owner are grouped under that owner’s best-ranked project.
- Members come from matching rules (GitHub topics, keywords in the description and README) plus a short hand-kept include and exclude list. The ranking is recomputed every hour from public GitHub activity.
Ranking
43 reposRanked by momentum, then health
- Rank 1
trycua/cuaOpen trycua/cua on GitHub (opens in new tab)
Scale computer-use 2.0 with open-source drivers, cross-OS sandboxes, and benchmarks for training, evaluation, and data generation.
+5.2k stars in 30 days +45% vs previous 30 days Score 97 - Rank 2
CursorTouch/Windows-MCPOpen CursorTouch/Windows-MCP on GitHub (opens in new tab)
MCP Server for Computer Use in Windows
+2.7k stars in 30 days +830% vs previous 30 days Score 96 - Rank 3
Tencent/BrowserSkillOpen Tencent/BrowserSkill on GitHub (opens in new tab)
Let AI agents use your real, logged-in browser without interrupting your work. CLI + extension for browser automation across any shell-capable AI agent.
+3.2k stars in 30 days −35% vs previous 30 days Score 93 - Rank 4
reticlehq/reticleOpen reticlehq/reticle on GitHub (opens in new tab)
AI agents can generate code, but still struggle to understand what they build. Reticle brings Jev-style machine-native runtime perception to web & desktop applications.
+507 stars in 30 days +17% vs previous 30 days Score 87 - Rank 5
vibheksoni/stealth-browser-mcpOpen vibheksoni/stealth-browser-mcp on GitHub (opens in new tab)
Stealth browser automation for AI agents over MCP and the Chrome DevTools Protocol: navigation, network hooks, DOM extraction, and pixel-accurate UI cloning.
+1k stars in 30 days +95% vs previous 30 days Score 85 - Rank 6
AtomicBot-ai/atomic-agentOpen AtomicBot-ai/atomic-agent on GitHub (opens in new tab)
Atomic Agent is a local-first AI agent. Runs open-weight models on your own machine via llama.cpp.
+1.1k stars in 30 days +275% vs previous 30 days Score 84 - Rank 7
leeguooooo/iphone-useOpen leeguooooo/iphone-use on GitHub (opens in new tab)
响应快 省Token, Let AI agents (Claude Code, Codex, any MCP client) drive a real iPhone · 让 AI agent 操作真实 iPhone:读屏、点击、输入、滑动,USB 或 Wi-Fi,自有 XCTest runner,可回放的 App flow。Open source, self-hosted on macOS.
+269 stars in 30 days +1,316% vs previous 30 days Score 83 - Rank 8
webbrain-one/webbrainOpen webbrain-one/webbrain on GitHub (opens in new tab)
Open-source AI browser agent for Chrome and Firefox (monorepo) 🧠
+413 stars in 30 days +61% vs previous 30 days Score 83 - Rank 9
GCWing/OpenBitFunOpen GCWing/OpenBitFun on GitHub (opens in new tab)
OpenBitFun combines a high-performance agent runtime written in Rust with a polished desktop application. It pairs the depth of a Code Agent with open, general-purpose capabilities for work beyond software development.
+568 stars in 30 days +10% vs previous 30 days Score 80 - Rank 10
lightpanda-io/browserOpen lightpanda-io/browser on GitHub (opens in new tab)
Lightpanda: the headless browser designed for AI and automation
+1.4k stars in 30 days +3% vs previous 30 days Score 78
- Rank 11
citrolabs/ego-liteOpen citrolabs/ego-lite on GitHub (opens in new tab)
The fastest browser for AI agents to run browser automation, built for sharing your logged-in browser state with your AI agents, like Codex or Claude Code, without disturbing you. Zero cost, zero config.
+1.9k stars in 30 days −50% vs previous 30 days Score 78 - Rank 12
browser-use/browser-useOpen browser-use/browser-use on GitHub (opens in new tab)
Open-source Python library for AI browser agents, plus Browser Use Cloud: stealth browsers and a hosted agent API.
+3.5k stars in 30 days −39% vs previous 30 days Score 76Also from browser-use
browser-use/video-useOpen browser-use/video-use on GitHub (opens in new tab)
Edit videos with coding agents
+4.3k stars in 30 days +4% vs previous 30 days Score 67browser-use/browser-harnessOpen browser-use/browser-harness on GitHub (opens in new tab)
Let Claude Code, Codex or Cursor use your real Chrome, logged in as you. One prompt to set up. Open source, MIT.
+702 stars in 30 days −27% vs previous 30 days Score 60browser-use/web-uiOpen browser-use/web-ui on GitHub (opens in new tab)
🖥️ Run AI Agent in your browser.
+87 stars in 30 days −72% vs previous 30 days Score 17 - Rank 13
iFurySt/open-computer-useOpen iFurySt/open-computer-use on GitHub (opens in new tab)
👾 Open Computer Use – Open-Source Alternative to Codex Computer Use
+413 stars in 30 days −3% vs previous 30 days Score 73 - Rank 14
vercel-labs/agent-browserOpen vercel-labs/agent-browser on GitHub (opens in new tab)
Browser automation CLI for AI agents
+1.5k stars in 30 days −22% vs previous 30 days Score 71 - Rank 15
jo-inc/camofox-browserOpen jo-inc/camofox-browser on GitHub (opens in new tab)
Stealth headless browser for AI agents — bypass Cloudflare, bot detection, and anti-scraping. Drop-in Puppeteer/Playwright replacement.
+671 stars in 30 days −72% vs previous 30 days Score 71 - Rank 16
cuga-project/cuga-agentOpen cuga-project/cuga-agent on GitHub (opens in new tab)
CUGA is an open-source generalist agent harness for the enterprise, supporting complex task execution on web and APIs, OpenAPI/MCP integrations, composable architecture, reasoning modes, and policy-aware features.
+181 stars in 30 days +1,545% vs previous 30 days Score 69 - Rank 17
microsoft/playwright-mcpOpen microsoft/playwright-mcp on GitHub (opens in new tab)
Playwright MCP server
+1.1k stars in 30 days −5% vs previous 30 days Score 68 - Rank 18
browserbase/stagehandOpen browserbase/stagehand on GitHub (opens in new tab)
The SDK to extract data and interact with any site on the web. Get started with Claude Code, Codex, Eve, Mastra, and more.
+1.1k stars in 30 days +57% vs previous 30 days Score 68 - Rank 19
CloakHQ/CloakBrowserOpen CloakHQ/CloakBrowser on GitHub (opens in new tab)
Stealth Chromium that passes every bot detection test. Drop-in Playwright replacement with source-level fingerprint patches. 50+ tests passed.
+896 stars in 30 days −25% vs previous 30 days Score 65 - Rank 20
getmaxun/maxunOpen getmaxun/maxun on GitHub (opens in new tab)
Turn any website into a structured API. Extract, automate, search and monitor the web.
+479 stars in 30 days +69% vs previous 30 days Score 65 - Rank 21
web-infra-dev/midsceneOpen web-infra-dev/midscene on GitHub (opens in new tab)
See interfaces like a human. E2E testing in natural language.
+417 stars in 30 days +30% vs previous 30 days Score 58 - Rank 22
testdriverai/testdriveraiOpen testdriverai/testdriverai on GitHub (opens in new tab)
Computer-Use SDK for E2E QA Testing
+232 stars in 30 days +5,700% vs previous 30 days Score 54 - Rank 23
browseros-ai/BrowserOSOpen browseros-ai/BrowserOS on GitHub (opens in new tab)
🌐 The open-source Agentic browser; alternative to ChatGPT Atlas, Perplexity Comet, Dia.
+351 stars in 30 days −28% vs previous 30 days Score 54 - Rank 24
Skyvern-AI/skyvernOpen Skyvern-AI/skyvern on GitHub (opens in new tab)
Automate browser based workflows with AI
+231 stars in 30 days −12% vs previous 30 days Score 49 - Rank 25
steel-dev/steel-browserOpen steel-dev/steel-browser on GitHub (opens in new tab)
🔥 Open Source Browser API for AI Agents & Apps. Steel Browser is a batteries-included browser sandbox that lets you automate the web without worrying about infrastructure.
+197 stars in 30 days +21% vs previous 30 days Score 46 - Rank 26
unbrowse-ai/unbrowseOpen unbrowse-ai/unbrowse on GitHub (opens in new tab)
Unbrowse - Reverse engineer any website so that your agents can access its APIs directly
+44 stars in 30 days +238% vs previous 30 days Score 44 - Rank 27
the-open-agent/openagentOpen the-open-agent/openagent on GitHub (opens in new tab)
⚡️next-generation personal AI assistant powered by LLM, RAG and agent loops, supporting computer-use, browser-use and coding agent, demo: https://demo.openagentai.org
+89 stars in 30 days −17% vs previous 30 days Score 43 - Rank 28
simular-ai/Agent-SOpen simular-ai/Agent-S on GitHub (opens in new tab)
Agent S: an open agentic framework that uses computers like a human
+360 stars in 30 days +112% vs previous 30 days Score 43 - Rank 29
LambdaTest/kane-cliOpen LambdaTest/kane-cli on GitHub (opens in new tab)
Kane CLI by TestMu AI (Formerly LambdaTest)
+12 stars in 30 days +200% vs previous 30 days Score 42 - Rank 30
lsdefine/GenericAgentOpen lsdefine/GenericAgent on GitHub (opens in new tab)
Self-evolving agent: grows skill tree from 3.3K-line seed, achieving full system control with 6x less token consumption
+164 stars in 30 days −56% vs previous 30 days Score 42 - Rank 31
nanobrowser/nanobrowserOpen nanobrowser/nanobrowser on GitHub (opens in new tab)
Open-source AI agent for web automation. Use your real, logged-in browser to automate daily tasks.
+358 stars in 30 days +38% vs previous 30 days Score 40 - Rank 32
CoderLuii/HolyClaudeOpen CoderLuii/HolyClaude on GitHub (opens in new tab)
AI coding workstation: Claude Code + web UI + 8 AI CLIs + headless browser + 50+ tools
+27 stars in 30 days −57% vs previous 30 days Score 37 - Rank 33
bytedance/UI-TARS-desktopOpen bytedance/UI-TARS-desktop on GitHub (opens in new tab)
The Open-Source Multimodal AI Agent Stack: Connecting Cutting-Edge AI Models and Agent Infra
+387 stars in 30 days −15% vs previous 30 days Score 35 - Rank 34
peters/horizonOpen peters/horizon on GitHub (opens in new tab)
GPU-accelerated terminal board that puts all your sessions on an infinite canvas
+13 stars in 30 days −7% vs previous 30 days Score 35 - Rank 35
tinyfish-io/tinyfish-cookbookOpen tinyfish-io/tinyfish-cookbook on GitHub (opens in new tab)
A collection of sample apps and recipes built with the TinyFish web agent. Open-source examples for you to learn & build!
+88 stars in 30 days +38% vs previous 30 days Score 35 - Rank 36
VyomKulshrestha/Heliox-OSOpen VyomKulshrestha/Heliox-OS on GitHub (opens in new tab)
Open-source, local-first desktop AI agent for voice, gesture, gaze, browser and system automation with permission gates and verification.
+2 stars in 30 days −33% vs previous 30 days Score 34 - Rank 37
apache/groovy-gebOpen apache/groovy-geb on GitHub (opens in new tab)
Apache Geb: Very Groovy Browser Automation
+0 stars in 30 days −100% vs previous 30 days Score 32 - Rank 38
go-vgo/robotgoOpen go-vgo/robotgo on GitHub (opens in new tab)
RobotGo, Go Native cross-platform RPA, GUI automation, Auto test and Computer use @vcaesar
+72 stars in 30 days +80% vs previous 30 days Score 30 - Rank 39
A9T9/RPAOpen A9T9/RPA on GitHub (opens in new tab)
Browser and desktop automation with an MCP server for AI assistants. Record macros or write JavaScript to automate websites, extract data, and control desktop apps with OCR and image recognition.
+63 stars in 30 days +37% vs previous 30 days Score 29 - Rank 40
vectora-foundry/native-devtools-mcpOpen vectora-foundry/native-devtools-mcp on GitHub (opens in new tab)
MCP server for computer use & browser automation - screenshot, OCR, click, type, find_text, Chrome/Electron CDP, template matching. macOS, Windows & Android. Works with Claude, Cursor, and any MCP client.
+4 stars in 30 days −43% vs previous 30 days Score 20
Showing 10 of 40 ranked repos. 22 newcomers are listed separately. 3 repos are folded under their owner’s best-ranked project.