Browser automation CLI for AI agents
★ 51,544⑂ 3,601👁 113⑃ 1,713● 807▤ 122
More from this author
+430
Daily
+3,521
Weekly
+7,774
Monthly
51,544
stars
3,601
forks
113
watchers
1,713
prs
807
issues
122
releases
README
agent-browser Browser automation for AI agents: open pages, read them as an accessibility tree, and act on elements from the shell. agent-browser is a native Rust CLI that gives an AI agent a browser it can drive one command at a time. snapshot returns the page as compact text with a ref such as @e1 for each element. The agent acts on a ref, reads the page again to check the result, and repeats. The browser stays open between commands, so each step is one short shell call. Install bash npm install -g agent-browser agent-browser install download Chrome for Testing (first time only) - Homebrew: brew install agent-browser, then agent-browser install. - Linux: agent-browser install --with-deps also installs the browser's system libraries. - No install: npx agent-browser open example.com. Project installs, building from source, upgrade, and doctor are covered in Installation. Quick start bash agent-browser open example.com agent-browser snapshot -i interactive elements, each with a ref agent-browser click @e1 act on a ref from that snapshot agent-browser get url check the result agent-browser close The snapshot for example.com: - link "Learn more" [ref=e1] - Refs describe the page at the time of the snapshot. After navigation or a DOM change, take a new snapshot before using refs again. - CSS selectors, text, XPath, and semantic locators also work, for example agent-browser find role button click --name "Submit". See Selectors. - A click fails, instead of reporting success, when the target is hidden, disabled, outside the viewport, or covered by another element such as a consent banner. Dismiss the covering element, take a new snapshot, and retry. The full walkthrough is in the quick start. How it works The CLI sends each command to a background daemon that starts on the first command and controls the browser over the Chrome DevTools Protocol. Each --session gets its own daemon and browser. By default, the daemon closes a headless browser and exits after one hour without commands, saving state first when session persistence is configured. Chrome for Testing is the default browser. You can also attach to a running Chrome with --cdp or --auto-connect (CDP mode), select the Lightpanda or Obscura engine with --engine, drive Mobile Safari in the iOS Simulator, or use a cloud browser provider such as Browserbase, Browserless, Browser Use, Kernel, or AgentCore. For AI agents The CLI ships skills that match its version: bash agent-browser skills get core the core loop, refs, waiting, and common workflows agent-browser skills get core --ref commands one reference, such as the full command list agent-browser skills get core --full adds the full command reference and templates agent-browser skills list specialized skills: electron, slack, dogfood, and more Install the discovery skill into an agent harness from skills.sh: bash npx
Top issues
#1247Background Mode to prevent stealing focus when connected to already-running Chrome instance10 comments · 2026-04-16#382Windows binary quarantined by Windows Defender — needs code signing6 comments · 2026-02-06#1068[Feature Request] CDP BrowserContext support for cookie-isolated parallel sessions5 comments · 2026-03-29#1713No per-command deadline: a wedged session target hangs snapshot/read/open indefinitely and harness-side timeouts cannot reap the detached daemon + Chrome tree4 comments · 2026-08-23#1406Chrome fails to launch when running agent-browser from an elevated shell on Windows4 comments · 2026-06-01#2121back/forward report "Inspected target navigated or closed" after the navigation succeeded3 comments · 2026-10-15#1973Recording aborts with encoder backlog at 1920x1080 and leaves zero-byte output3 comments · 2026-09-22#1854macOS: headless launches insert a spurious Google Chrome tile in the Dock3 comments · 2026-09-15
Discussions
#2605Accept saved state under --allowed-domains when every site in it is allowed0 comments · 2026-11-14#2604Optional preconditions for actions: exact refs, an expected element, and browser-reported password fields1 comments · 2026-11-14#2588Could provide portable version on Windows ?1 comments · 2026-11-13#2525Dashboard: let embedding tools limit engines, providers and the Chat tab1 comments · 2026-11-10#2519A two-state invoice regression: fresh snapshots, model abstention, and a disabled-click check0 comments · 2026-11-10#2517help0 comments · 2026-11-10#2364Publish a schema for --json output and MCP tool results (outputSchema)1 comments · 2026-11-03#2366Feature request: configurable paths and ports for daemon, Chrome CDP and stream server (for OS-sandboxed use)0 comments · 2026-11-03
Latest release
v0.43.0 v0.43.0 2026-11-12New Features - Added snapshot hints for controls whose accessible name is empty or shared with another ref in the same snapshot. The hint follows the ref and lists the control's icon, image file, data-testid, name or id, and the text of the nearest row or list item (2578) - Added a Modern Web Guidance skill that retrieves Chrome's guidance for a task and checks the built page in agent-browser (2577) Bug Fixes - Fixed commands on closed tabs waiting for the timeout. A command whose tab closes before Chrome answers now fails with Target closed: CDP session detached (2576) - Fixed recordings
相关报道 (7)
QQ News浅谈 AI 进化史2026-09-27QQ News你的Hermes,正在使用你的身份……2026-08-28yage.aiAgent 的浏览器正在分裂成两极:真实态与并发密度为什么不可兼得2026-08-14QQ News实测ego lite,给我Codex浏览器自动化加到2.5倍速了!2026-07-3080aj.com独立Agent工具落地:开发者开源GenericAgent的浏览器技能库2026-04-1551CTO安装完这16个Skills,我的Openclaw变聪明了2026-03-09品玩再进化!网易有道 LobsterAI 发布重大更新,首推 Skill Store 与 MCP 协议支持2026-03-06
News & coverage (10)
Hacker NewsShow HN: Skill that lets Claude Code/Codex spin up VMs and GPUs▲ 138 · 36 comments · 2026-02-13Hacker NewsShow HN: Smooth CLI – Token-efficient browser for AI agents▲ 109 · 74 comments · 2026-02-05Hacker NewsAgent-Browser by Vercel Labs▲ 2 · 0 comments · 2026-01-25Hacker NewsAgent-browser by Vercel: Browser automation CLI for AI agents▲ 2 · 0 comments · 2026-01-13Hacker NewsHeadless browser automation CLI for AI agents from Vercel▲ 2 · 0 comments · 2026-01-12markets.businessinsider.comAgentR launches Webcmd, open-source browser infrastructure that lets AI agents learn a website once2026-10-09The GitHub BlogYour contributors are AI-first now. Is your project?2026-08-12TechGigVercel’s Rust-Based Agent-Browser Targets AI Automation Workflows2026-05-15Visual Studio MagazineVS Code 1.119 Adds Agent Browser Sharing, OpenTelemetry Tracing2026-05-07AIBaseAI Can Finally Manipulate Things! Vercel Launches Agent Browser to Let Large Models Control Websites2026-01-13