REA
Reverse engineer anything from your terminal or agent with one CLI and MCP server.
Let Claude Code, Codex or Cursor use your real Chrome, logged in as you. One prompt to set up. Open source, MIT.
제작자라면 이 코드를 README에 붙여 넣으세요. 배지는 이 페이지로 연결됩니다.
[](https://mcp.popnx.com/en/skills/browser-use-browser-harness)같은 작업에서 AI를 서비스에 연결하는 MCP
Reverse engineer anything from your terminal or agent with one CLI and MCP server.
Codebase intelligence for TypeScript and JavaScript, exposed as structured MCP analysis tools.
MCP server for terminal commands, file operations, and process management
같은 종류의 작업, Skill 목록과 같은 순서
| 이 항목 browser-harness | rea | andrej-karpathy-skills | graphify | MoneyPrinterTurbo | |
|---|---|---|---|---|---|
| 제작자 | browser-use | morluto | multica-ai | Graphify-Labs | harry0703 |
| Skill 수 | 1 | 1 | 1 | 1 | 1 |
| GitHub 별 | 19,098 | 463,670 | 233,682 | 148,887 | 143,987 |
| 마지막 코드 변경 | 2026년 11월 16일 | 2026년 11월 17일 | 2026년 4월 20일 | 2026년 11월 17일 | 2026년 11월 16일 |
| 라이선스 | MIT | MIT | – | Apache-2.0 | MIT |
browser-use의 Skill(지시 묶음) 1개, AI가 읽고 그대로 일해요 · MCP server가 아니에요
MCP = 손
AI를 외부 서비스나 데이터에 연결합니다(예: Notion 읽기, 청구서 만들기). 프로그램이나 서버가 실행 중이어야 합니다.
Skill = 안내서
작업 방법을 AI에게 가르치는 텍스트 파일. 어떤 서비스에도 연결되지 않습니다.
겹치는 경우도 있습니다. 도구를 부르는 스크립트가 있는 Skill도, 지식만 보내는 MCP도 있습니다. 하려는 작업을 기준으로 고르세요.
Claude CodeCodexCursorGemini CLIGitHub Copilot
제작자 README와 repo 폴더 기준
GitHub account browser-use (organization), 1 years old
★ 19,098 stars · code last changed 2026년 11월 16일
36 scripts that can run on a computer: browser-harness, agent-workspace/agent_helpers.py, agent-workspace/domain-skills/browser-use-cloud/cleanup-zombies.py, src/browser_harness/__init__.py and 32 more
Scripts that go online: tests/unit/test_skill.py, tests/unit/test_brave_origin.py
No hooks or writes to machine settings found
Checked automatically from the repo files at commit 1894799 (read only, nothing was run). Not a safety guarantee: read it and decide for yourself
먼저: 어떤 AI를 쓰시나요
쓰는 앱을 고르면 그 앱의 설치 방법만 보여 드립니다
Files are byte-for-byte the same as on GitHub at commit 1894799 · View on GitHub ↗
36 scripts
Once uploaded it runs on Claude’s systems, not on your computer. No need to unzip
All 241 files · 4.5 MB
.claude-plugin/marketplace.json1 KB.claude-plugin/plugin.json1 KB.env.example1 KBAGENTS.md2 KBCONTRIBUTING.md1 KBLICENSE1 KBREADME.md9 KBSKILL.md14 KBagent-workspace/agent_helpers.py1 KBagent-workspace/domain-skills/BOSS-zhipin/chat.md13 KBagent-workspace/domain-skills/BOSS-zhipin/job-search.md13 KBagent-workspace/domain-skills/BOSS-zhipin/navigation.md3 KBagent-workspace/domain-skills/admin/microsoft-365.md1 KBagent-workspace/domain-skills/agentlist/discovery.md6 KBagent-workspace/domain-skills/alaska/checkout.md6 KBagent-workspace/domain-skills/ap/namecheap.md1 KBagent-workspace/domain-skills/apollo/search-and-lists.md5 KBagent-workspace/domain-skills/app/apollo.md1 KBagent-workspace/domain-skills/app/smartlead.md1 KBagent-workspace/domain-skills/archive-org/scraping.md15 KBagent-workspace/domain-skills/art-institute-chicago/scraping.md15 KBagent-workspace/domain-skills/articulate-rise/code-blocks.md9 KBagent-workspace/domain-skills/arxiv-bulk/scraping.md14 KBagent-workspace/domain-skills/arxiv/scraping.md13 KBagent-workspace/domain-skills/atlas/overview.md3 KBagent-workspace/domain-skills/aviationweather/scraping.md30 KBagent-workspace/domain-skills/bank-of-canada/scraping.md32 KBagent-workspace/domain-skills/bbb/scraping.md24 KBagent-workspace/domain-skills/bible-api/scraping.md13 KBagent-workspace/domain-skills/bigbang-hr/checkout.md5 KBagent-workspace/domain-skills/bilbasen/search.md4 KBagent-workspace/domain-skills/bilibili/navigation.md11 KBagent-workspace/domain-skills/bluesky/scraping.md35 KBagent-workspace/domain-skills/booking-com/scraping.md22 KBagent-workspace/domain-skills/booking/frontier-checkout.md1 KBagent-workspace/domain-skills/browser-use-cloud/cleanup-zombies.py6 KBagent-workspace/domain-skills/browser-use-cloud/cloud.md9 KBagent-workspace/domain-skills/capterra/scraping.md18 KBagent-workspace/domain-skills/carbon-intensity/scraping.md25 KBagent-workspace/domain-skills/centilebrain/generate-estimates.md5 KBagent-workspace/domain-skills/clinicaltrials-gov/scraping.md29 KBagent-workspace/domain-skills/coingecko/scraping.md14 KBagent-workspace/domain-skills/coinmarketcap/scraping.md17 KBagent-workspace/domain-skills/costco/scraping.md25 KBagent-workspace/domain-skills/craigslist/scraping.md15 KBagent-workspace/domain-skills/crossref/scraping.md23 KBagent-workspace/domain-skills/ctrip/hotels.md10 KBagent-workspace/domain-skills/dashboard/stripe.md1 KBagent-workspace/domain-skills/defillama/scraping.md33 KBagent-workspace/domain-skills/dev-to/scraping.md13 KBagent-workspace/domain-skills/dexscreener/scraping.md45 KBagent-workspace/domain-skills/docker-hub/scraping.md33 KBagent-workspace/domain-skills/docs/google-sheets.md1 KBagent-workspace/domain-skills/duckduckgo/scraping.md15 KBagent-workspace/domain-skills/eurostat/scraping.md31 KBagent-workspace/domain-skills/eventbrite/scraping.md15 KBagent-workspace/domain-skills/expedia/api.md3 KBagent-workspace/domain-skills/expedia/automation.md5 KBagent-workspace/domain-skills/expedia/flights-search.md10 KBagent-workspace/domain-skills/federal-register/scraping.md24 KBagent-workspace/domain-skills/feishu/docx-scraping.md5 KBagent-workspace/domain-skills/finviz/scraping.md21 KBagent-workspace/domain-skills/flighthub/flights-search.md7 KBagent-workspace/domain-skills/flightradar24/scraping.md20 KBagent-workspace/domain-skills/flipkart/shopping.md10 KBagent-workspace/domain-skills/flyfrontier/checkout.md1 KBagent-workspace/domain-skills/fotmob/scraping.md21 KBagent-workspace/domain-skills/framer/editor.md7 KBagent-workspace/domain-skills/frankfurter/scraping.md23 KBagent-workspace/domain-skills/free-dictionary/scraping.md15 KBagent-workspace/domain-skills/frontier/checkout.md7 KBagent-workspace/domain-skills/gbif/scraping.md35 KBagent-workspace/domain-skills/gemini/spark-skills.md2 KBagent-workspace/domain-skills/github/repo-actions.md3 KBagent-workspace/domain-skills/github/scraping.md8 KBagent-workspace/domain-skills/gmail/compose.md5 KBagent-workspace/domain-skills/goodreads/scraping.md19 KBagent-workspace/domain-skills/google-maps/scraping.md21 KBagent-workspace/domain-skills/google-patents/scraping.md27 KBagent-workspace/domain-skills/google-play/scraping.md32 KBagent-workspace/domain-skills/google-sheets/writing-cells.md3 KBagent-workspace/domain-skills/google-trends/scraping.md21 KBagent-workspace/domain-skills/google/flights-multi-city.md5 KBagent-workspace/domain-skills/hackernews/scraping.md8 KBagent-workspace/domain-skills/home-depot/scraping.md43 KBagent-workspace/domain-skills/hubspot/private-app-webhooks.md6 KBagent-workspace/domain-skills/huggingface/scraping.md32 KBagent-workspace/domain-skills/iherb/product-labels.md3 KBagent-workspace/domain-skills/imdb/scraping.md12 KBagent-workspace/domain-skills/itch-io/scraping.md17 KBagent-workspace/domain-skills/jetblue/checkout.md8 KBagent-workspace/domain-skills/lichess/scraping.md18 KBagent-workspace/domain-skills/loom/folder-enumeration.md9 KBagent-workspace/domain-skills/loom/read-recording.md5 KBagent-workspace/domain-skills/lulu/account-currency.md2 KBagent-workspace/domain-skills/manus/tasks.md12 KBagent-workspace/domain-skills/mercari/scraping.md25 KBagent-workspace/domain-skills/mermaid/render.md3 KBagent-workspace/domain-skills/met-museum/scraping.md16 KBagent-workspace/domain-skills/microsoft-365/domain-setup.md2 KBagent-workspace/domain-skills/musicbrainz/scraping.md19 KBagent-workspace/domain-skills/namecheap/advanced-dns.md2 KBagent-workspace/domain-skills/nasa/scraping.md12 KBagent-workspace/domain-skills/nhtsa/scraping.md23 KBagent-workspace/domain-skills/noaa/scraping.md17 KBagent-workspace/domain-skills/nobel-prize/scraping.md21 KBagent-workspace/domain-skills/nvd/scraping.md25 KBagent-workspace/domain-skills/ods/fact-sheets.md2 KBagent-workspace/domain-skills/oecd/scraping.md30 KBagent-workspace/domain-skills/open-library/scraping.md19 KBagent-workspace/domain-skills/open-trivia/scraping.md21 KBagent-workspace/domain-skills/openalex/scraping.md18 KBagent-workspace/domain-skills/openfda/scraping.md27 KBagent-workspace/domain-skills/openstreetmap/scraping.md20 KBagent-workspace/domain-skills/ourworldindata/scraping.md28 KBagent-workspace/domain-skills/outlook/webmail-compose.md2 KBagent-workspace/domain-skills/package-registries/npm-pypi.md16 KBagent-workspace/domain-skills/partsouq/vin-lookup.md2 KBagent-workspace/domain-skills/perplexity/computer.md16 KBagent-workspace/domain-skills/pokeapi/scraping.md31 KBagent-workspace/domain-skills/police-uk/scraping.md26 KBagent-workspace/domain-skills/polymarket/scraping.md13 KBagent-workspace/domain-skills/poshmark/scraping.md25 KBagent-workspace/domain-skills/producthunt/scraping.md11 KBagent-workspace/domain-skills/pubmed/scraping.md18 KBagent-workspace/domain-skills/qbo/report-export.md6 KBagent-workspace/domain-skills/reddit/scraping.md8 KBagent-workspace/domain-skills/redfin/scraping.md33 KBagent-workspace/domain-skills/rest-countries/scraping.md10 KBagent-workspace/domain-skills/rightmove/scraping.md33 KBagent-workspace/domain-skills/scryfall/scraping.md30 KBagent-workspace/domain-skills/sec-edgar/scraping.md16 KBagent-workspace/domain-skills/shopify-admin/README.md2 KBagent-workspace/domain-skills/shopify-admin/embedded-apps.md3 KBagent-workspace/domain-skills/shopify-admin/forms-app.md2 KBagent-workspace/domain-skills/shopify-admin/knowledge-base.md4 KBagent-workspace/domain-skills/shopify-admin/polaris-inputs.md5 KBagent-workspace/domain-skills/shopify-storefronts/scraping.md25 KBagent-workspace/domain-skills/slack/files.md3 KBagent-workspace/domain-skills/sleeper/scraping.md38 KBagent-workspace/domain-skills/smartlead/mailbox-warmup.md3 KBagent-workspace/domain-skills/sofascore/scraping.md25 KBagent-workspace/domain-skills/steam/scraping.md22 KBagent-workspace/domain-skills/stripe/account-sandboxes.md2 KBagent-workspace/domain-skills/substack/scraping.md12 KBagent-workspace/domain-skills/supabase/dashboard.md3 KBagent-workspace/domain-skills/swapi/scraping.md16 KBagent-workspace/domain-skills/swiss-transport/scraping.md30 KBagent-workspace/domain-skills/tasksquad-ai/agents.md3 KBagent-workspace/domain-skills/tasksquad-ai/auth.md3 KBagent-workspace/domain-skills/tasksquad-ai/tasks.md6 KBagent-workspace/domain-skills/tcgplayer/scraping.md23 KBagent-workspace/domain-skills/tfl/scraping.md30 KBagent-workspace/domain-skills/thetechgeeks/pricing.md3 KBagent-workspace/domain-skills/thorne/product-labels.md2 KBagent-workspace/domain-skills/ticketmaster/scraping.md22 KBagent-workspace/domain-skills/treasury-fiscal-data/scraping.md46 KBagent-workspace/domain-skills/trello/boards-and-lists.md4 KBagent-workspace/domain-skills/trulia/scraping.md25 KBagent-workspace/domain-skills/tvmaze/scraping.md40 KBagent-workspace/domain-skills/twitch/scraping.md41 KBagent-workspace/domain-skills/uniprot/scraping.md34 KBagent-workspace/domain-skills/upwork/scraping.md19 KBagent-workspace/domain-skills/usgs-earthquake/scraping.md22 KBagent-workspace/domain-skills/vercel/archive-and-audit.md3 KBagent-workspace/domain-skills/vercel/observability.md2 KBagent-workspace/domain-skills/vercel/vercel.md7 KBagent-workspace/domain-skills/wayback-machine/scraping.md11 KBagent-workspace/domain-skills/weather/scraping.md17 KBagent-workspace/domain-skills/wehotel/hotels.md10 KBagent-workspace/domain-skills/wellfound/scraping.md21 KBagent-workspace/domain-skills/weread/read.md9 KBagent-workspace/domain-skills/wikidata/scraping.md40 KBagent-workspace/domain-skills/wikipedia/scraping.md28 KBagent-workspace/domain-skills/world-bank/scraping.md12 KBagent-workspace/domain-skills/x/article-source-recovery.md2 KBagent-workspace/domain-skills/x/posting.md3 KBagent-workspace/domain-skills/xiaohongshu/scraping.md6 KBagent-workspace/domain-skills/youtube/playback.md4 KBagent-workspace/domain-skills/youtube/scraping.md18 KBbrowser-harness1 KBdocs/MCP.md2 KBdocs/allow-remote-debugging.png188 KBdocs/download-latest-20-x-videos.gif1.3 MBdocs/setup-remote-debugging.png50 KBdocs/snap-linux-headless.md2 KBinstall.md4 KBinteraction-skills/connection.md2 KBinteraction-skills/cookies.md1 KBinteraction-skills/cross-origin-iframes.md1 KBinteraction-skills/dialogs.md2 KBinteraction-skills/downloads.md1 KBinteraction-skills/drag-and-drop.md1 KBinteraction-skills/dropdowns.md1 KBinteraction-skills/iframes.md1 KBinteraction-skills/make-video.md4 KBinteraction-skills/network-requests.md1 KBinteraction-skills/print-as-pdf.md1 KBinteraction-skills/profile-sync.md5 KBinteraction-skills/screenshots.md1 KBinteraction-skills/scrolling.md1 KBinteraction-skills/shadow-dom.md1 KBinteraction-skills/tabs.md2 KBinteraction-skills/uploads.md1 KBinteraction-skills/viewport.md1 KBpyproject.toml1 KBskills/browser-harness/references/install.md3 KBsrc/browser_harness/__init__.py1 KBsrc/browser_harness/_ipc.py9 KBsrc/browser_harness/admin.py65 KBsrc/browser_harness/auth.py18 KBsrc/browser_harness/daemon.py40 KBsrc/browser_harness/helpers.py27 KBsrc/browser_harness/macos.py4 KBsrc/browser_harness/mcp_cli.py1 KBsrc/browser_harness/paths.py1 KBsrc/browser_harness/recorder.py12 KBsrc/browser_harness/run.py14 KBsrc/browser_harness/telemetry.py9 KBsrc/browser_harness/video-template.html69 KBsrc/browser_harness/video.py29 KBsrc/browser_harness/video_render.py20 KBsrc/mcp_server.py8 KBtests/__init__.py1 KBtests/conftest.py1 KBtests/integration/__init__.py1 KBtests/integration/test_js.py6 KBtests/unit/__init__.py1 KBtests/unit/test_admin.py60 KBtests/unit/test_auth.py1 KBtests/unit/test_brave_origin.py3 KBtests/unit/test_daemon.py31 KBtests/unit/test_helpers.py28 KBtests/unit/test_ipc.py5 KBtests/unit/test_macos.py4 KBtests/unit/test_mcp_cli.py1 KBtests/unit/test_mcp_server.py1 KBtests/unit/test_recorder.py3 KBtests/unit/test_run.py14 KBtests/unit/test_skill.py1 KBtests/unit/test_video_render.py1 KB+Create skillUpload a skillpick browser-harness.zip
SettingsCapabilitiesCode execution and file creation
Turn on Code execution and file creation first (newer menus: Create and edit files) · personal plans: Settings → Capabilities · Team/Enterprise: an admin turns it on under Organization settings → Plugins & skills · every plan, including Free
Upload once; it works in claude.ai and Claude Desktop, and in recent Claude Code signed in with a Claude account (not with an API key)
Cowork and cloud sessions use the skills turned on in your claude.ai account, not ~/.claude/skills on your computer (Claude Desktop scheduled tasks run locally and still load it)
Checked against official docs 2026년 11월 12일 · Source: support.claude.com, support.claude.com, code.claude.com
이 문장을 복사해 Claude Code 채팅에 붙여 넣으세요
For one project, paste it into a chat that is open in that project’s folder · The AI always shows you the path before installing
The AI tells you what it will do first and asks your permission at every step
Claude Code를 다시 열고 /browser-harness를 입력하거나 관련 작업을 요청하면 Claude가 알아서 써요
Checked against official docs 2026년 11월 12일 · Source: code.claude.com
Skill은 AI가 따르는 지시예요. 설치 전에 파일을 읽고, 믿을 수 있는 제작자 것만 설치하세요.
Mac: Command+스페이스 후 Terminal 입력 · Windows: 아래 명령은 PowerShell에서 안 돼요. 위의 “AI에게 설치 맡기기”를 쓰세요
repo를 임시 폴더에 내려받고 SKILL.md가 있는 폴더만 복사해요
git clone --depth 1 https://github.com/browser-use/browser-harness.git /tmp/popnx-skill-browser-use-browser-harness-1894799 && mkdir -p ~/.claude/skills && cp -R '/tmp/popnx-skill-browser-use-browser-harness-1894799/.' ~/.claude/skills/browser-harnessChatGPT 데스크톱 앱에는 Codex가 들어 있고, Codex는 ~/.agents/skills 폴더에서 Skill을 읽습니다
ChatGPT 데스크톱 앱에서 Codex를 열고 이 문장을 채팅에 붙여 넣으세요
The AI always shows you the path before installing
The AI tells you what it will do first and asks your permission at every step
browser-harness 폴더를 ~/.agents/skills/browser-harness에 넣으세요
~는 홈 폴더($HOME)입니다
터미널 없이: ZIP을 내려받아 압축을 풀고 browser-harness 폴더를 ~/.agents/skills로 옮기세요
Skill은 AI가 따르는 지시예요. 설치 전에 파일을 읽고, 믿을 수 있는 제작자 것만 설치하세요.
Mac: Command+스페이스 후 Terminal 입력 · Windows: 아래 명령은 PowerShell에서 안 돼요. 위의 “AI에게 설치 맡기기”를 쓰세요
repo를 임시 폴더에 내려받고 SKILL.md가 있는 폴더만 복사해요
git clone --depth 1 https://github.com/browser-use/browser-harness.git /tmp/popnx-skill-browser-use-browser-harness-1894799 && mkdir -p ~/.agents/skills && cp -R '/tmp/popnx-skill-browser-use-browser-harness-1894799/.' ~/.agents/skills/browser-harnessChecked against official docs 2026년 11월 15일 · Source: learn.chatgpt.com
직접 설치한 Skill은 쓸 수 없고, 디렉터리의 플러그인에 들어 있는 Skill만 쓸 수 있습니다
이 Skill을 ChatGPT에서 쓰려면 위 방법대로 ChatGPT 데스크톱 앱을 쓰세요
Checked against official docs 2026년 11월 15일 · Source: learn.chatgpt.com
이 문장을 복사해 Codex 채팅에 붙여 넣으세요 · ChatGPT 데스크톱 앱 안의 Codex, Codex CLI, IDE 확장 프로그램에서 모두 쓸 수 있습니다
For one project, paste it into a chat that is open in that project’s folder · The AI always shows you the path before installing
The AI tells you what it will do first and asks your permission at every step
Codex는 새로 설치한 Skill을 스스로 찾습니다. 보이지 않으면 Codex를 다시 시작하세요 · 쓰는 법: ChatGPT에서는 @를 입력하고 Skill을 고르세요 · Codex CLI나 IDE 확장 프로그램에서는 /skills를 실행하거나 $ 뒤에 Skill 이름을 입력하세요
Checked against official docs 2026년 11월 15일 · Source: learn.chatgpt.com
Skill은 AI가 따르는 지시예요. 설치 전에 파일을 읽고, 믿을 수 있는 제작자 것만 설치하세요.
Mac: Command+스페이스 후 Terminal 입력 · Windows: 아래 명령은 PowerShell에서 안 돼요. 위의 “AI에게 설치 맡기기”를 쓰세요
repo를 임시 폴더에 내려받고 SKILL.md가 있는 폴더만 복사해요
git clone --depth 1 https://github.com/browser-use/browser-harness.git /tmp/popnx-skill-browser-use-browser-harness-1894799 && mkdir -p ~/.agents/skills && cp -R '/tmp/popnx-skill-browser-use-browser-harness-1894799/.' ~/.agents/skills/browser-harness이 문장을 복사해 Cursor 채팅에 붙여 넣으세요
For one project, paste it into a chat that is open in that project’s folder · The AI always shows you the path before installing
The AI tells you what it will do first and asks your permission at every step
Cursor Agent 채팅에서 /를 입력하고 browser-harness 선택
Checked against official docs 2026년 11월 12일 · Source: cursor.com
Skill은 AI가 따르는 지시예요. 설치 전에 파일을 읽고, 믿을 수 있는 제작자 것만 설치하세요.
Mac: Command+스페이스 후 Terminal 입력 · Windows: 아래 명령은 PowerShell에서 안 돼요. 위의 “AI에게 설치 맡기기”를 쓰세요
repo를 임시 폴더에 내려받고 SKILL.md가 있는 폴더만 복사해요
git clone --depth 1 https://github.com/browser-use/browser-harness.git /tmp/popnx-skill-browser-use-browser-harness-1894799 && mkdir -p ~/.cursor/skills && cp -R '/tmp/popnx-skill-browser-use-browser-harness-1894799/.' ~/.cursor/skills/browser-harnessSkill을 읽을 수 있는 다른 앱용입니다. 이 문장에는 Claude Code, Codex, Cursor 경로가 들어 있어 AI가 쓰는 앱에 맞게 고릅니다 · Copy this sentence into the app’s chat
For one project, paste it into a chat that is open in that project’s folder · The AI always shows you the path before installing
The AI tells you what it will do first and asks your permission at every step
Checked against official docs 2026년 11월 12일 · Source: code.claude.com, learn.chatgpt.com, cursor.com
Gemini CLI, GitHub Copilot 같은 다른 Agent Skills 지원 앱은 폴더가 달라요. 그 앱 안내를 보고 SKILL.md가 있는 폴더를 복사하세요.
2026년 11월 11일에 Claude Code, Codex, Cursor 공식 안내로 skill 폴더 확인
✓ 명령 확인 2026년 11월 16일
아래 글은 제작자 repo에서 온 것이며 PopNx MCP의 지시가 아니에요
처음 8,000자 표시 · GitHub에서 전체 파일 열기
---
name: browser-harness
description: "Control a real browser via CDP: clicking, typing, navigation, logged-in sessions, JS-rendered or bot-protected pages. Not for plain HTTP fetches of public content - use curl for those."
---
# browser-harness
Direct browser control via CDP. For task-specific edits, use `agent-workspace/agent_helpers.py`. For setup, install, or connection problems, read https://github.com/browser-use/browser-harness/blob/main/install.md.
## When Not to Use
A basic fetch of public information needs no browser. If a plain HTTP request can read it — a public page, an API, docs — use `curl` or your fetch tool, and leave the browser alone. Use browser-harness when the task needs interaction (click, type, navigate), the user's logged-in session, JS rendering, or a bot-protected page. If a direct fetch fails or returns a shell page, then escalate to the browser.
Domain skills are off by default. Set `BH_DOMAIN_SKILLS=1` to enable them; see the bottom section.
**If `BH_DOMAIN_SKILLS=1` and the task is site-specific, read every file in the matching `$BH_AGENT_WORKSPACE/domain-skills/<site>/` directory before inventing an approach.**
## Usage
```bash
browser-harness <<'PY'
print(page_info())
PY
```
- Invoke as `browser-harness`. Use heredocs for multi-line commands.
- Helpers are pre-imported. `run.py` calls `ensure_daemon()` before `exec`.
- First navigation for a task is `new_tab(url)`, not `goto_url(url)`. The daemon
preserves the attached tab across separate CLI invocations, so do not call
`new_tab()` again in every script.
- Keep one working tab per task/site. Before opening another, inspect
`current_tab()` and `list_tabs()` and use `switch_tab()` to reuse a matching
tab. Do not leave duplicate tabs on the same URL or close tabs you did not
create.
- At task completion, close tabs created for the task that are no longer needed.
Keep a tab open if the user needs to see it, it is needed for a known follow-up,
or closing it could discard unsaved work or other important state.
- `new_tab()` and `switch_tab()` attach and move the horse marker without
changing Chrome's visible tab. Screenshots and normal CDP input work in the
background. Never call `activate_tab(target)` automatically: it brings Chrome
to the foreground. Call it only when the user explicitly asks to see or
visibly switch to that tab. Do not pair `switch_tab()` with `activate_tab()`.
- A local daemon is a connection to the whole Chrome instance, not to one site,
task, card, or agent. Omit `BU_NAME` and reuse the default daemon for normal
sequential local work across websites, tabs, screenshots, and Codex turns.
Do not invent per-job names such as `gmail1375` or `slack1371`: every new
local daemon opens another browser-level CDP connection and Chrome may show
another Allow prompt.
- Set `BH_TAB_MARKER=0` before starting the daemon to leave page titles unchanged.
The horse marker remains enabled by default.
- A timeout or page that pauses while hidden is not permission to foreground
Chrome. Keep using background CDP operations. For a focus-gated page,
temporarily call `cdp("Emulation.setFocusEmulationEnabled", enabled=True)`,
perform and verify the operation, then disable it in a `finally` block. If
background control still cannot work, report that limitation instead of
activating the tab. Do not invent a `Runtime.evaluate` scroll replacement or
a cross-frame JS walker.
- The normal local flow attaches to the running Chrome/Chromium CDP endpoint. No browser ids or local profile selection.
## Local Chrome
The default daemon can keep many tabs and visit many sites; browser-harness has
no per-site, screenshot, or result-count limit that requires a new daemon.
Chrome memory and page complexity are the practical limits. Reuse matching tabs
with `list_tabs()` and `switch_tab()`.
One daemon has one mutable attached/current tab. Many agents can share it when
their browser operations are serialized: treat local Chrome as one shared
browser lane while non-browser work continues in parallel. Sequential tab
switching, input, and screenshot capture are safe. Do not create another local
daemon merely because several agents exist.
Two agents that switch tabs and act simultaneously can race, causing one to act
on or capture the other's tab. For truly simultaneous interactive work, use
separate remote browsers when Browser Use Cloud authentication is already
available. Otherwise serialize browser operations through the default local
daemon. A named local daemon is a last resort when simultaneous isolation is
required, remote auth is unavailable or unsuitable, and the extra Chrome
approval prompt is acceptable. It creates another controller and dedicated tab
in the same local Chrome profile, not another Chrome profile or process.
If the default daemon becomes stale, use its built-in reattachment/recovery
first. A command timeout, truncated output, site change, closed tab, or new task
is not a reason to create another daemon. Run `browser-harness --doctor` and
restart or replace the default daemon only when it is actually dead or cannot
recover.
If the daemon cannot connect, run diagnostics:
```bash
browser-harness --doctor
```
If Chrome is not running at all, the harness launches it automatically and retries.
If Chrome is running but remote debugging is not enabled, the harness opens:
```text
chrome://inspect/#remote-debugging
```
On macOS, when local Chrome asks for remote-debugging permission, keep the
original browser command running and call `mac-approve` in another shell/tool
call. Preserve the exact daemon name: if the waiting command used
`BU_NAME=r7k2`, run:
```text
BU_NAME=r7k2 browser-harness mac-approve
```
For the default daemon, omit the `BU_NAME` prefix. The original command resumes
when the helper returns `ready`; do not rerun it. If the helper reports
`accessibility-required`, ask the user once to grant the app launching
browser-harness (for example Terminal, iTerm, or Codex) access in System
Settings > Privacy & Security > Accessibility, then call `mac-approve` once
again. This is only for local Chrome; do not call it for `BU_CDP_URL`,
`BU_CDP_WS`, or Browser Use Cloud.
When the shell tool can yield a still-running process, use a short 3-5 second
initial yield for the first local command, not a 30-second wait. If the command
yields with the Allow hint, leave that exact process running, immediately call
`browser-harness mac-approve` in a second tool call, then resume or poll the
original process. With a named daemon, preserve its exact `BU_NAME` for the
helper. Never start the browser command again. If the user clicks Allow first,
the same handshake completes and the original command returning successfully
is the agent's feedback; `mac-approve` also returns `ready` when the daemon is
already connected.
`mac-approve` is macOS-only. On Linux or Windows, keep the original browser
command running and ask the user to click Allow if Chrome presents the approval
dialog. Their click completes the same handshake, so resume or poll the original
process for success; do not rerun it or create a replacement daemon. If that
Chrome build presents no approval dialog, the original command simply connects.
## Remote Browsers
Use Browser Use cloud for headless servers, parallel sub-agents, or isolated work.
Remote browsers require Browser Use Cloud authentication. Check
`browser-harness auth status` before depending on them. `browser-harness auth
login` stores authentication for later processes, so an API key does not need to
be passed to every agent process; without stored authentication or an available
`BROWSER_USE_API_KEY`, serialize work through the default local daemon instead.
Cloud browsers are managed Chrome instances hosted by Browser Use. Each one is a fresh, isolated browser. Proactively suggest one (briefly explain why) when:
- **The user wants multiple concurrent tasks.** Local Chrome is one shared browser; p질문하고, 팁을 나누고, 안 되는 점을 알려 주세요. "도움이 됨"이 많은 순서예요.