Browser Use
Browser agents and stealth cloud browsers for AI: run web tasks, fill forms, extract data.
作者による説明(英語の原文)
Drive a stealth-patched Chromium daemon: navigate, screenshot, eval JS, attach over CDP.
この3つを1つのスコアにまとめることはしません。意見、測定、人気はそれぞれ別のものだからです。
このサーバーのやさしい解説はまだ書いていません。以下はレジストリから読み取った事実です。
使っているAIアプリを選んで、設定をコピーしてください。
BROWSER_USE_API_KEY, BROWSER_USE_URL「秘密」と表示された値は、あなたのアカウントの鍵です。絶対に人に教えたり、チャットに貼り付けたりしないでください。
必要なもの
どのアプリを使っていますか?
上でアプリを選んでください。選んだアプリを覚えておくので、次からはボタンに「…にインストール」と表示されます。
Claude Desktop → Settings → Developer → Edit Config を開き、claude_desktop_config.json にこれを貼り付けます
{
"mcpServers": {
"sendblue-browser": {
"command": "npx",
"args": [
"-y",
"sendblue-browser-mcp"
],
"env": {
"BROWSER_USE_API_KEY": "your-value",
"BROWSER_USE_URL": "your-value"
}
}
}
}ターミナルでこれを実行します
claude mcp add sendblue-browser --env BROWSER_USE_API_KEY=your-value --env BROWSER_USE_URL=your-value -- npx -y sendblue-browser-mcp動かす前に、空欄の値を入力してください。
プロジェクトの .cursor/mcp.json、または ~/.cursor/mcp.json に貼り付けます
{
"mcpServers": {
"sendblue-browser": {
"command": "npx",
"args": [
"-y",
"sendblue-browser-mcp"
],
"env": {
"BROWSER_USE_API_KEY": "your-value",
"BROWSER_USE_URL": "your-value"
}
}
}
}ブラウザがアプリを開いてよいか確認し、そのあとアプリが設定を表示して確認を求めます。キーの値は空欄のままなので、自分で入力してください。
プロジェクトの .vscode/mcp.json に貼り付けます
{
"servers": {
"sendblue-browser": {
"type": "stdio",
"command": "npx",
"args": [
"-y",
"sendblue-browser-mcp"
],
"env": {
"BROWSER_USE_API_KEY": "your-value",
"BROWSER_USE_URL": "your-value"
}
}
}
}ブラウザがアプリを開いてよいか確認し、そのあとアプリが設定を表示して確認を求めます。キーの値は空欄のままなので、自分で入力してください。
これはパソコンで動くサーバーですが、ChatGPTがつなげるのはインターネット上のサーバーだけです。ほかのアプリを選んでください。
選んだアプリは、この端末のCookieに保存されます。アカウントは不要です。
BROWSER_USE_URL envこの設定は、作者のレジストリ登録情報から作成しています。READMEと違う場合は、READMEに従ってください。
パソコンで動くサーバーです。インストールすると、ツール一覧がAIアプリに表示されます。
誰でも見られる情報をチェックしており、コードは読んでいません。当てはまらない項目は数に入れていません。 · 2026年11月7日にチェック
レジストリ名は、リポジトリを所有するGitHubアカウント sendblue-api で確認済みです
このリポジトリはまだGitHubから読み取っていません
このリポジトリはまだGitHubから読み取っていません
これはパソコンで動くため、インターネット上のエンドポイントはありません
パソコンで動き、代わりにローカルの設定を使います
パソコンで動くサーバーです。ツール一覧を読み取れるのは、インターネット上のエンドポイントがある場合だけです
レジストリでこのカテゴリにある有効なすべてのサーバーと比較しています。スターは、公開者がリポジトリの所有者である場合だけ数えます。
当てはまる基本チェック1項目のうち1項目に合格しました。コードは読んでいません。キーやトークンを追加する前に、サーバーがどんなアクセスを求めているかを確認し、必要な権限だけを持つキーを使ってください。
入力が必要なもの:BROWSER_USE_API_KEY
あなたのパソコンで動きます。必要なときにAIアプリが起動し、データは作者のサーバーを通りません(サーバー自体がオンラインサービスを呼び出す場合を除きます)。
GitHubリポジトリの最終更新は–、最新バージョン(0.2.3)がレジストリに届いたのは5 か月前です。
Claude Desktop、Claude Code、Cursor、VS CodeなどのMCP対応アプリで使えます。パソコンにNode.jsも必要です。
同じカテゴリには:Browser Use, Chrome DevTools MCP, MCP Server Browser
同じカテゴリで、ほかの作者が作ったサーバー。スターの多い順。
カードの見方:★ GitHubスターi3つの柱:レビュー · チェック · 利用状況i更新=最新バージョンをレジストリに公開した日i
Browser agents and stealth cloud browsers for AI: run web tasks, fill forms, extract data.
MCP server for Chrome DevTools
MCP server for browser use access
作者のリポジトリから書かれたとおりにコピーしたものです。PopNx MCPは内容を確認・推奨していません。HTML、画像、スクリプトは取り除いています。リンクはこのサイトの外へ移動します。
A standalone debug browser for agents. Stealth-patched Chromium behind a tiny HTTP API, with persistent named sessions, an easy purge endpoint, configurable auto screenshots, and CDP attach so any client (Playwright, Puppeteer, undetected-chromedriver, custom scripts) can drive it.
It runs as its own process — not tied to any one Claude / Cursor / Codex session — so multiple agents can share it concurrently, debug each other's flows, and reuse a logged-in session without paying the auth tax every run.
| Need | What you get |
|---|---|
| Not coupled to a single agent | Long-running HTTP daemon. Any tool with a bearer token can drive it. |
| Doesn't look automated | Chromium patched via patchright: navigator.webdriver is hidden, CDP detection vectors are patched, real Chromium binary (no Playwright fingerprints). Passes typical Cloudflare Turnstile and similar low-friction checks; not for hostile scraping at scale. |
| Reusable sessions | Named persistent profiles. Log in once, every subsequent debug run reuses the cookies. |
| One-click reset | POST /sessions/:name/purge clears cookies + active-page storage but keeps the session id, so your client code keeps working. |
| Multi-agent friendly | Each session is an isolated BrowserContext (or its own profile). Run 5+ debug sessions in parallel without state bleed. |
| CDP attach | GET /sessions/:name/cdp-url returns the shared browser CDP URL for non-persistent sessions, so Playwright / Puppeteer / any CDP client can connect() to the debug browser. |
| Auto evidence | Navigation screenshots follow NAV_SCREENSHOT_POLICY (default: headless only). Optional Playwright traces. |
| HTTP-controlled | Plain curl works. No SDK required. |
| Self-contained deploy | One docker compose up on any Mac mini / EC2 / Codespace. No host Chromium needed. |
Prereqs: Bun >=1.3 and Node.js >=20 on your PATH.
git clone https://github.com/sendblue-api/sendblue-browser-use.git
cd sendblue-browser-use bun install bun --bun x patchright install chromium
cp .env.example .env
TOKEN="$(openssl rand -hex 32)" sed -i.bak "s/^BROWSER_USE_API_KEY=.*/BROWSER_USE_API_KEY=$TOKEN/" .env rm .env.bak
bun run dev
Then in another terminal:
source .env
TOKEN="Authorization: Bearer $BROWSER_USE_API_KEY"
curl -s -X POST http://127.0.0.1:8787/sessions \ -H "$TOKEN" -H "Content-Type: application/json" \ -d '{"name":"qa","persistent":true}'
curl -s -X POST http://127.0.0.1:8787/sessions/qa/navigate \ -H "$TOKEN" -H "Content-Type: application/json" \ -d '{"url":"https://trybloom.so/"}'
curl -s http://127.0.0.1:8787/sessions/qa/screenshot?fullPage=true \ -H "$TOKEN" -o /tmp/qa.png
curl -s -X POST http://127.0.0.1:8787/sessions/qa/purge -H "$TOKEN"
echo "BROWSER_USE_API_KEY=$(openssl rand -hex 32)" > .env
docker compose up -d
Same API, exposed on 127.0.0.1:8787. Data persists in ./data/. CDP stays loopback-only inside the container by default; to attach CDP from the host, uncomment CDP_BIND and the 127.0.0.1:9222:9222 port mapping in docker-compose.yml on a trusted machine only.
The repo ships three packaging layers so any agent can drive the daemon with one line of config.
The cross-agent path. Wraps the HTTP API as MCP tools (health, create_session, navigate, screenshot, script, get_cdp_url, purge_session, etc).
{
"mcpServers": { "sendblue-browser": { "command": "npx", "args": ["-y", "sendblue-browser-mcp@0.2.3"], "env": { "BROWSER_USE_URL": "http://127.0.0.1:8787", "BROWSER_USE_API_KEY": " " } } } }
Drop into ~/.cursor/mcp.json, ~/Library/Application Support/Claude/claude_desktop_config.json, ~/.gemini/config/mcp_config.json, etc. For Codex:
codex mcp add sendblue-browser \
--env BROWSER_USE_URL=http://127.0.0.1:8787 \ --env BROWSER_USE_API_KEY= \ -- npx -y sendblue-browser-mcp@0.2.3
Wrapper source: mcp/.
/plugin install sendblue-api/sendblue-browser-use
Pulls .claude-plugin/plugin.json + skills/sendblue-browser/SKILL.md from this repo.
The same skills/sendblue-browser/SKILL.md works as a skill across Codex, Google Antigravity, and Cline. Copy the folder into:
| Agent | Path |
|---|---|
| Codex | ~/.codex/skills/sendblue-browser/ |
| Antigravity | ~/.gemini/skills/sendblue-browser/ |
| Cline | enable Skills in Settings, then drop in its skills directory |
If your repo already follows the agents.md convention, the AGENTS.md in this repo doubles as a copy-pasteable section for any consumer repo that wants to call this daemon.
All routes require Authorization: Bearer $BROWSER_USE_API_KEY except /health.
| Method | Path | Body / Query | Returns |
|---|---|---|---|
GET | /health | — | { ok, service, version, sessions, defaultHeadless, navScreenshotPolicy } |
GET | /sessions | — | { sessions: [...] } |
POST | /sessions | { name, persistent?, headless?, viewport?, userAgent?, locale?, timezone?, traces?, proxy? } | { session } |
GET | /sessions/:name | — | session info + current page url/title |
POST | /sessions/:name/purge | — | { session } — clear cookies + active-origin storage, keep session id |
DELETE | /sessions/:name | — | { ok } — close context, remove profile |
POST | /sessions/:name/navigate | { url, waitUntil?, timeoutMs? } | { url, title, status } |
GET | /sessions/:name/screenshot | ?fullPage=true&selector=... | image/png |
POST | /sessions/:name/script | { code, timeoutMs? } (≤200kB) | { result } — code runs in page context |
GET | /sessions/:name/cookies | ?url=... | { cookies } |
POST | /sessions/:name/cookies | { cookies: [...] } | { ok } |
GET | /sessions/:name/console | ?limit=100 | { messages: [...] } (ring buffer) |
GET | /sessions/:name/cdp-url | — | { cdpUrl, targetId } — for non-persistent sessions only |
POST /sessions/:name/script passes code directly to Playwright's page.evaluate. Send a JavaScript expression, or wrap statements in an IIFE. Scripts time out after 30s by default; pass timeoutMs up to 120000, or 0 to disable the request timeout.
{ "code": "(() => { const x = 1; return x; })()" }
All non-2xx responses share the same envelope. Failures from Playwright are truncated to one line before being returned; the full message is logged server-side with proxy credentials redacted.
{ "error": { "code": "navigate_failed", "message": "net::ERR_NAME_NOT_RESOLVED" } }
| Status | Code | Meaning |
|---|---|---|
| 400 | empty_body | request body is required but was missing |
| 400 | malformed_json | body was not valid JSON |
| 400 | invalid_body | body failed Zod validation (see error.details) |
| 401 | unauthorized | missing or wrong bearer token |
| 404 | not_found | session does not exist |
| 409 | already_exists | a session with that name is already running |
| 409 | cdp_unavailable_for_persistent_session | persistent sessions don't expose the shared CDP url |
| 422 | script_failed | /script request was valid but the in-page eval threw or timed out |
| 500 | internal_error | unexpected — check server logs |
| 502 | navigate_failed screenshot_failed cookie_read_failed cookie_write_failed session_unreachable | Chromium-side failure |
{
name: string; // required, [a-z0-9_-] persistent?: boolean; // default true — profile data survives restarts headless?: boolean | "new"; // persistent sessions only; non-persistent uses DEFAULT_HEADLESS viewport?: { width, height }; userAgent?: string; locale?: string; // default "en-US" timezone?: string; traces?: boolean; // capture a Playwright trace.zip on session close proxy?: { server, username?, password?, bypass? }; }
POST /sessions/:name/cookies accepts up to 500 Playwright-style cookies:
{
cookies: Array<{ name: string; value: string; url?: string; // absolute URL, or domain + path domain?: string; path?: string; expires?: number; httpOnly?: boolean; secure?: boolean; sameSite?: "Strict" | "Lax" | "None"; }>; }
Each cookie must include either url or both domain and path.
~/.sendblue-browser-use/profiles/ /. After a daemon restart, recreate the same named persistent session to reuse that profile. CDP attach is not available for these sessions because each gets its own private browser instance.BrowserContext. State is in-memory only. CDP attach IS available via /sessions/:name/cdp-url, which returns the shared browser-level CDP endpoint plus a targetId for the session page. CDP clients can see the shared debug browser, so keep the CDP port local and trusted and select the returned target before driving.Pick persistent for "log in once, debug for a week" workflows. Pick non-persistent when you need to attach Playwright / Puppeteer directly.
examples/attach-from-playwright.ts — connect Playwright over CDP and drive a page (run with Node/tsx as shown in the file header)examples/attach-from-puppeteer.ts — same with Puppeteerexamples/multi-agent-debug.ts — two agents driving two surfaces in parallelexamples/run-script.sh — curl-only walkthroughThis is built on patchright, a maintained fork of Playwright that:
navigator.webdriverWe deliberately do not pass --disable-blink-features=AutomationControlled or override userAgent/viewport by default — patchright handles those internally and our overrides would defeat its rebrowser patches. Pass an explicit userAgent per session only if you have a reason.
Why patchright over alternatives:
~/.sendblue-browser-use/
├── profiles/ │ └── / # persistent profile dir └── runs/ └── / ├── 2026-05-25T...-nav.png # automatic nav screenshot when policy enables it └── 2026-05-25T...-trace.zip # if traces:true at create
POST /sessions/:name/purge clears cookies and permissions context-wide, then clears localStorage, sessionStorage, IndexedDB, ServiceWorker registrations, CacheStorage, and the console buffer for c
冒頭だけを表示しています。続きはリポジトリで読めます