[{"_1":2,"_655":-5,"_656":-5},"loaderData",{"_3":4,"_29":30,"_31":32,"_489":490},"root",{"_5":6,"_14":15},"toasts",[7],{"_8":9,"_10":11,"_12":13},"uid","01971fa5-7b28-702c-8ee3-ed39a0ae4252","title","This MCP server does not support inspection","type","error","visitor",{"_16":17,"_18":19},"viewportWidth",1280,"visitorSession",{"_20":21,"_23":24,"_25":-5,"_26":27,"_28":-5},"attributes",[22],"BOT","id",7058884,"membership","referenceId","01971406-43e3-74e8-b880-2f000258007a","userAccount","routes/_public/_layout",{},"routes/_public/mcp/servers/~namespace/~slug/_pages/_layout",{"_33":34,"_89":-5,"_90":47,"_36":91,"_407":408,"_488":224},"availableLocales",[35,47,54,61,68,75,82],{"_36":37,"_40":44,"_45":46},"mcpServer",{"_38":39,"_40":41,"_42":43},"description","Ein Model Context Protocol-Server, der es LLMs ermöglicht, über strukturierte Zugänglichkeits-Snapshots mit Webseiten zu interagieren, ohne dass Vision-Modelle oder Screenshots erforderlich sind.","name","Playwright MCP-Server","readme","## Dramatiker MCP\n\nEin Model Context Protocol (MCP)-Server, der Browser-Automatisierungsfunktionen mit [Playwright](https://playwright.dev) bereitstellt. Dieser Server ermöglicht LLMs die Interaktion mit Webseiten über strukturierte Zugänglichkeits-Snapshots, sodass keine Screenshots oder visuell optimierten Modelle erforderlich sind.\n\n### Hauptmerkmale\n\n* **Schnell und leicht** . Verwendet den Barrierefreiheitsbaum von Playwright, keine pixelbasierte Eingabe.\n* **LLM-freundlich** . Keine Vision-Modelle erforderlich, arbeitet ausschließlich mit strukturierten Daten.\n* **Deterministische Toolanwendung** . Vermeidet Mehrdeutigkeiten, die bei Screenshot-basierten Ansätzen häufig auftreten.\n\n### Anforderungen\n\n* Node.js 18 oder neuer\n* VS Code, Cursor, Windsurf, Claude Desktop oder jeder andere MCP-Client\n\n### Erste Schritte\n\nInstallieren Sie zunächst den Playwright MCP-Server mit Ihrem Client. Eine typische Konfiguration sieht wie folgt aus:\n\n```js\n{\n \"mcpServers\": {\n \"playwright\": {\n \"command\": \"npx\",\n \"args\": [\n \"@playwright/mcp@latest\"\n ]\n }\n }\n}\n```\n\nSie können den Playwright MCP-Server auch mithilfe der VS Code CLI installieren:\n\n```bash\n# For VS Code\ncode --add-mcp '{\"name\":\"playwright\",\"command\":\"npx\",\"args\":[\"@playwright/mcp@latest\"]}'\n```\n\nNach der Installation steht der Playwright MCP-Server zur Verwendung mit Ihrem GitHub Copilot-Agent in VS Code zur Verfügung.\n\nGehen Sie zu `Cursor Settings` -> `MCP` -> `Add new MCP Server` . Benennen Sie den Server nach Ihren Wünschen und verwenden Sie den `command` mit dem Befehl `npx @playwright/mcp` . Sie können die Konfiguration auch überprüfen oder befehlsähnliche Argumente hinzufügen, indem Sie auf `Edit` klicken.\n\n```js\n{\n \"mcpServers\": {\n \"playwright\": {\n \"command\": \"npx\",\n \"args\": [\n \"@playwright/mcp@latest\"\n ]\n }\n }\n}\n```\n\nBefolgen Sie die Windsuff MCP- [Dokumentation](https://docs.windsurf.com/windsurf/cascade/mcp) . Verwenden Sie die folgende Konfiguration:\n\n```js\n{\n \"mcpServers\": {\n \"playwright\": {\n \"command\": \"npx\",\n \"args\": [\n \"@playwright/mcp@latest\"\n ]\n }\n }\n}\n```\n\nFolgen Sie der MCP- [Installationsanleitung](https://modelcontextprotocol.io/quickstart/user) und verwenden Sie die folgende Konfiguration:\n\n```js\n{\n \"mcpServers\": {\n \"playwright\": {\n \"command\": \"npx\",\n \"args\": [\n \"@playwright/mcp@latest\"\n ]\n }\n }\n}\n```\n\n### Konfiguration\n\nDer Playwright MCP-Server unterstützt folgende Argumente. Sie können in der obigen JSON-Konfiguration als Teil der `\"args\"` -Liste angegeben werden:\n\n```\n> npx @playwright/mcp@latest --help\n --allowed-origins semicolon-separated list of origins to allow the\n browser to request. Default is to allow all.\n --blocked-origins semicolon-separated list of origins to block the\n browser from requesting. Blocklist is evaluated\n before allowlist. If used without the allowlist,\n requests not matching the blocklist are still\n allowed.\n --block-service-workers block service workers\n --browser browser or chrome channel to use, possible\n values: chrome, firefox, webkit, msedge.\n --caps comma-separated list of capabilities to enable,\n possible values: tabs, pdf, history, wait, files,\n install. Default is all.\n --cdp-endpoint CDP endpoint to connect to.\n --config path to the configuration file.\n --device device to emulate, for example: \"iPhone 15\"\n --executable-path path to the browser executable.\n --headless run browser in headless mode, headed by default\n --host host to bind server to. Default is localhost. Use\n 0.0.0.0 to bind to all interfaces.\n --ignore-https-errors ignore https errors\n --isolated keep the browser profile in memory, do not save\n it to disk.\n --no-image-responses do not send image responses to the client.\n --no-sandbox disable the sandbox for all process types that\n are normally sandboxed.\n --output-dir path to the directory for output files.\n --port port to listen on for SSE transport.\n --proxy-bypass comma-separated domains to bypass proxy, for\n example \".com,chromium.org,.domain.com\"\n --proxy-server specify proxy server, for example\n \"http://myproxy:3128\" or \"socks5://myproxy:8080\"\n --save-trace Whether to save the Playwright Trace of the\n session into the output directory.\n --storage-state path to the storage state file for isolated\n sessions.\n --user-agent specify user agent string\n --user-data-dir path to the user data directory. If not\n specified, a temporary directory will be created.\n --viewport-size specify browser viewport size in pixels, for\n example \"1280, 720\"\n --vision Run server that uses screenshots (Aria snapshots\n are used by default)\n```\n\n### Benutzerprofil\n\nSie können Playwright MCP mit einem persistenten Profil wie einen normalen Browser (Standard) oder in isolierten Kontexten für die Testsitzungen ausführen.\n\n**Persistentes Profil**\n\nAlle Anmeldeinformationen werden im persistenten Profil gespeichert. Sie können es zwischen den Sitzungen löschen, wenn Sie den Offline-Status aufheben möchten. Das persistente Profil befindet sich an den folgenden Speicherorten und kann mit dem Argument `--user-data-dir` überschrieben werden.\n\n```bash\n# Windows\n%USERPROFILE%\\AppData\\Local\\ms-playwright\\mcp-{channel}-profile\n\n# macOS\n- ~/Library/Caches/ms-playwright/mcp-{channel}-profile\n\n# Linux\n- ~/.cache/ms-playwright/mcp-{channel}-profile\n```\n\n**Isoliert**\n\nIm isolierten Modus wird jede Sitzung im isolierten Profil gestartet. Jedes Mal, wenn Sie MCP auffordern, den Browser zu schließen, wird die Sitzung beendet und der gesamte Speicherstatus für diese Sitzung geht verloren. Sie können dem Browser den anfänglichen Speicherstatus über die `contextOptions` der Konfiguration oder über das Argument `--storage-state` bereitstellen. Weitere Informationen zum Speicherstatus finden Sie [hier](https://playwright.dev/docs/auth) .\n\n```js\n{\n \"mcpServers\": {\n \"playwright\": {\n \"command\": \"npx\",\n \"args\": [\n \"@playwright/mcp@latest\",\n \"--isolated\",\n \"--storage-state={path/to/storage.json}\n ]\n }\n }\n}\n```\n\n### Konfigurationsdatei\n\nDer Playwright MCP-Server kann mithilfe einer JSON-Konfigurationsdatei konfiguriert werden. Sie können die Konfigurationsdatei mit der Befehlszeilenoption `--config` angeben:\n\n```bash\nnpx @playwright/mcp@latest --config path/to/config.json\n```\n\n```typescript\n{\n // Browser configuration\n browser?: {\n // Browser type to use (chromium, firefox, or webkit)\n browserName?: 'chromium' | 'firefox' | 'webkit';\n\n // Keep the browser profile in memory, do not save it to disk.\n isolated?: boolean;\n\n // Path to user data directory for browser profile persistence\n userDataDir?: string;\n\n // Browser launch options (see Playwright docs)\n // @see https://playwright.dev/docs/api/class-browsertype#browser-type-launch\n launchOptions?: {\n channel?: string; // Browser channel (e.g. 'chrome')\n headless?: boolean; // Run in headless mode\n executablePath?: string; // Path to browser executable\n // ... other Playwright launch options\n };\n\n // Browser context options\n // @see https://playwright.dev/docs/api/class-browser#browser-new-context\n contextOptions?: {\n viewport?: { width: number, height: number };\n // ... other Playwright context options\n };\n\n // CDP endpoint for connecting to existing browser\n cdpEndpoint?: string;\n\n // Remote Playwright server endpoint\n remoteEndpoint?: string;\n },\n\n // Server configuration\n server?: {\n port?: number; // Port to listen on\n host?: string; // Host to bind to (default: localhost)\n },\n\n // List of enabled capabilities\n capabilities?: Array<\n 'core' | // Core browser automation\n 'tabs' | // Tab management\n 'pdf' | // PDF generation\n 'history' | // Browser history\n 'wait' | // Wait utilities\n 'files' | // File handling\n 'install' | // Browser installation\n 'testing' // Testing\n >;\n\n // Enable vision mode (screenshots instead of accessibility snapshots)\n vision?: boolean;\n\n // Directory for output files\n outputDir?: string;\n\n // Network configuration\n network?: {\n // List of origins to allow the browser to request. Default is to allow all. Origins matching both `allowedOrigins` and `blockedOrigins` will be blocked.\n allowedOrigins?: string[];\n\n // List of origins to block the browser to request. Origins matching both `allowedOrigins` and `blockedOrigins` will be blocked.\n blockedOrigins?: string[];\n };\n \n /**\n * Do not send image responses to the client.\n */\n noImageResponses?: boolean;\n}\n```\n\n### Eigenständiger MCP-Server\n\nWenn Sie einen Headed Browser auf einem System ohne Anzeige oder aus Arbeitsprozessen der IDEs ausführen, führen Sie den MCP-Server aus der Umgebung mit DISPLAY aus und übergeben Sie das Flag `--port` um den SSE-Transport zu aktivieren.\n\n```bash\nnpx @playwright/mcp@latest --port 8931\n```\n\nUnd dann legen Sie in der MCP-Clientkonfiguration die `url` zum SSE-Endpunkt fest:\n\n```js\n{\n \"mcpServers\": {\n \"playwright\": {\n \"url\": \"http://localhost:8931/sse\"\n }\n }\n}\n```\n\n**HINWEIS:** Die Docker-Implementierung unterstützt derzeit nur Headless-Chromium.\n\n```js\n{\n \"mcpServers\": {\n \"playwright\": {\n \"command\": \"docker\",\n \"args\": [\"run\", \"-i\", \"--rm\", \"--init\", \"--pull=always\", \"mcr.microsoft.com/playwright/mcp\"]\n }\n }\n}\n```\n\nSie können das Docker-Image selbst erstellen.\n\n```\ndocker build -t mcr.microsoft.com/playwright/mcp .\n```\n\n```js\nimport http from 'http';\n\nimport { createConnection } from '@playwright/mcp';\nimport { SSEServerTransport } from '@modelcontextprotocol/sdk/server/sse.js';\n\nhttp.createServer(async (req, res) => {\n // ...\n\n // Creates a headless Playwright MCP server with SSE transport\n const connection = await createConnection({ browser: { launchOptions: { headless: true } } });\n const transport = new SSEServerTransport('/messages', res);\n await connection.connect(transport);\n\n // ...\n});\n```\n\n### Werkzeuge\n\nDie Tools sind in zwei Modi verfügbar:\n\n1. **Snapshot-Modus** (Standard): Verwendet Zugänglichkeits-Snapshots für bessere Leistung und Zuverlässigkeit\n2. **Vision-Modus** : Verwendet Screenshots für visuell basierte Interaktionen\n\nUm den Vision-Modus zu verwenden, fügen Sie beim Starten des Servers das Flag `--vision` hinzu:\n\n```js\n{\n \"mcpServers\": {\n \"playwright\": {\n \"command\": \"npx\",\n \"args\": [\n \"@playwright/mcp@latest\",\n \"--vision\"\n ]\n }\n }\n}\n```\n\nDer Vision-Modus funktioniert am besten mit Computermodellen, die in der Lage sind, mit Elementen im XY-Koordinatenraum zu interagieren, basierend auf dem bereitgestellten Screenshot.\n\n* **Browser-Schnappschuss**\n\n * Titel: Seiten-Schnappschuss\n * Beschreibung: Erfassen Sie einen Barrierefreiheits-Schnappschuss der aktuellen Seite. Dies ist besser als ein Screenshot.\n * Parameter: Keine\n * Schreibgeschützt: **true**\n\n* **Browserklick**\n\n * Titel: Klick\n * Beschreibung: Klicken Sie auf einer Webseite\n * Parameter:\n * `element` (Zeichenfolge): Für Menschen lesbare Elementbeschreibung, die verwendet wird, um die Berechtigung zur Interaktion mit dem Element zu erhalten\n * `ref` (Zeichenfolge): Genaue Zielelementreferenz aus dem Seiten-Snapshot\n * Schreibgeschützt: **false**\n\n* **browser\\_drag**\n\n * Titel: Maus ziehen\n * Beschreibung: Drag & Drop zwischen zwei Elementen durchführen\n * Parameter:\n * `startElement` (Zeichenfolge): Für Menschen lesbare Beschreibung des Quellelements, die verwendet wird, um die Berechtigung zur Interaktion mit dem Element zu erhalten\n * `startRef` (Zeichenfolge): Genaue Quellelementreferenz aus dem Seiten-Snapshot\n * `endElement` (Zeichenfolge): Für Menschen lesbare Zielelementbeschreibung, die verwendet wird, um die Berechtigung zur Interaktion mit dem Element zu erhalten\n * `endRef` (Zeichenfolge): Genaue Zielelementreferenz aus dem Seiten-Snapshot\n * Schreibgeschützt: **false**\n\n* **browser\\_hover**\n\n * Titel: Maus darüber schweben\n * Beschreibung: Bewegen Sie den Mauszeiger über das Element auf der Seite\n * Parameter:\n * `element` (Zeichenfolge): Für Menschen lesbare Elementbeschreibung, die verwendet wird, um die Berechtigung zur Interaktion mit dem Element zu erhalten\n * `ref` (Zeichenfolge): Genaue Zielelementreferenz aus dem Seiten-Snapshot\n * Schreibgeschützt: **true**\n\n* **Browsertyp**\n\n * Titel: Text eingeben\n * Beschreibung: Geben Sie Text in ein bearbeitbares Element ein\n * Parameter:\n * `element` (Zeichenfolge): Für Menschen lesbare Elementbeschreibung, die verwendet wird, um die Berechtigung zur Interaktion mit dem Element zu erhalten\n * `ref` (Zeichenfolge): Genaue Zielelementreferenz aus dem Seiten-Snapshot\n * `text` (Zeichenfolge): Text, der in das Element eingegeben werden soll\n * `submit` (Boolesch, optional): Ob eingegebener Text abgesendet werden soll (nachher die Eingabetaste drücken)\n * `slowly` (Boolesch, optional): Gibt an, ob jeweils ein Zeichen eingegeben werden soll. Nützlich zum Auslösen von Tastenhandlern auf der Seite. Standardmäßig wird der gesamte Text auf einmal ausgefüllt.\n * Schreibgeschützt: **false**\n\n* **Browserauswahloption**\n\n * Titel: Option auswählen\n * Beschreibung: Wählen Sie eine Option in einer Dropdown-Liste aus\n * Parameter:\n * `element` (Zeichenfolge): Für Menschen lesbare Elementbeschreibung, die verwendet wird, um die Berechtigung zur Interaktion mit dem Element zu erhalten\n * `ref` (Zeichenfolge): Genaue Zielelementreferenz aus dem Seiten-Snapshot\n * `values` (Array): Array von Werten, die in der Dropdown-Liste ausgewählt werden sollen. Dies kann ein einzelner oder mehrere Werte sein.\n * Schreibgeschützt: **false**\n\n* **Browser\\_Press\\_Key**\n\n * Titel: Drücken Sie eine Taste\n * Beschreibung: Drücken Sie eine Taste auf der Tastatur\n * Parameter:\n * `key` (string): Name der zu drückenden Taste oder eines zu generierenden Zeichens, wie etwa `ArrowLeft` oder `a`\n * Schreibgeschützt: **false**\n\n* **Browser\\_wait\\_for**\n\n * Titel: Warten auf\n * Beschreibung: Warten Sie, bis Text erscheint oder verschwindet oder eine bestimmte Zeit vergeht\n * Parameter:\n * `time` (Zahl, optional): Die Wartezeit in Sekunden\n * `text` (Zeichenfolge, optional): Der Text, auf den gewartet werden soll\n * `textGone` (Zeichenfolge, optional): Der Text, auf dessen Verschwinden gewartet werden soll\n * Schreibgeschützt: **true**\n\n* **Browser-Datei-Upload**\n\n * Titel: Dateien hochladen\n * Beschreibung: Eine oder mehrere Dateien hochladen\n * Parameter:\n * `paths` (Array): Die absoluten Pfade zu den hochzuladenden Dateien. Kann eine einzelne oder mehrere Dateien sein.\n * Schreibgeschützt: **false**\n\n* **Browser-Handle-Dialog**\n\n * Titel: Einen Dialog führen\n * Beschreibung: Behandeln eines Dialogs\n * Parameter:\n * `accept` (Boolesch): Ob der Dialog akzeptiert werden soll.\n * `promptText` (Zeichenfolge, optional): Der Text der Eingabeaufforderung im Falle eines Eingabeaufforderungsdialogs.\n * Schreibgeschützt: **false**\n\n* **browser\\_navigate**\n\n * Titel: Navigieren Sie zu einer URL\n * Beschreibung: Navigieren Sie zu einer URL\n * Parameter:\n * `url` (Zeichenfolge): Die URL, zu der navigiert werden soll\n * Schreibgeschützt: **false**\n\n* **browser\\_navigate\\_back**\n\n * Titel: Zurück\n * Beschreibung: Zurück zur vorherigen Seite\n * Parameter: Keine\n * Schreibgeschützt: **true**\n\n* **Browser\\_Navigate\\_Forward**\n\n * Titel: Weiter geht's\n * Beschreibung: Weiter zur nächsten Seite\n * Parameter: Keine\n * Schreibgeschützt: **true**\n\n* **Browser\\_Machen\\_Screenshot**\n\n * Titel: Machen Sie einen Screenshot\n * Beschreibung: Erstellen Sie einen Screenshot der aktuellen Seite. Sie können keine Aktionen basierend auf dem Screenshot ausführen. Verwenden Sie für Aktionen browser\\_snapshot.\n * Parameter:\n * `raw` (Boolesch, optional): Gibt an, ob ohne Komprimierung (im PNG-Format) zurückgegeben werden soll. Der Standardwert ist „false“, wodurch ein JPEG-Bild zurückgegeben wird.\n * `filename` (Zeichenfolge, optional): Dateiname, in dem der Screenshot gespeichert werden soll. Standardmäßig wird `page-{timestamp}.{png|jpeg}` verwendet, falls nicht anders angegeben.\n * `element` (Zeichenfolge, optional): Lesbare Elementbeschreibung, die verwendet wird, um die Berechtigung zum Erstellen eines Screenshots des Elements zu erhalten. Falls nicht angegeben, wird der Screenshot vom Ansichtsfenster erstellt. Falls ein Element angegeben ist, muss auch die Referenz angegeben werden.\n * `ref` (Zeichenfolge, optional): Genaue Zielelementreferenz aus dem Seiten-Snapshot. Falls nicht angegeben, wird der Screenshot des Ansichtsfensters erstellt. Wenn ref angegeben ist, muss auch das Element angegeben werden.\n * Schreibgeschützt: **true**\n\n* **browser\\_pdf\\_save**\n\n * Titel: Als PDF speichern\n * Beschreibung: Seite als PDF speichern\n * Parameter:\n * `filename` (Zeichenfolge, optional): Dateiname, in dem die PDF-Datei gespeichert werden soll. Der Standardwert ist `page-{timestamp}.pdf` wenn nicht anders angegeben.\n * Schreibgeschützt: **true**\n\n* **Browser-Netzwerkanforderungen**\n\n * Titel: Netzwerkanfragen auflisten\n * Beschreibung: Gibt alle Netzwerkanfragen seit dem Laden der Seite zurück\n * Parameter: Keine\n * Schreibgeschützt: **true**\n\n* **Browser-Konsolennachrichten**\n\n * Titel: Konsolenmeldungen abrufen\n * Beschreibung: Gibt alle Konsolenmeldungen zurück\n * Parameter: Keine\n * Schreibgeschützt: **true**\n\n* **Browserinstallation**\n\n * Titel: Installieren Sie den in der Konfiguration angegebenen Browser\n * Beschreibung: Installiert den in der Konfiguration angegebenen Browser. Rufen Sie diese Option auf, wenn eine Fehlermeldung angezeigt wird, dass der Browser nicht installiert ist.\n * Parameter: Keine\n * Schreibgeschützt: **false**\n\n* **browser\\_close**\n\n * Titel: Browser schließen\n * Beschreibung: Seite schließen\n * Parameter: Keine\n * Schreibgeschützt: **true**\n\n* **Browsergröße ändern**\n\n * Titel: Browserfenstergröße ändern\n * Beschreibung: Größe des Browserfensters ändern\n * Parameter:\n * `width` (Zahl): Breite des Browserfensters\n * `height` (Zahl): Höhe des Browserfensters\n * Schreibgeschützt: **true**\n\n* **Browser-Tab-Liste**\n\n * Titel: Listenregisterkarten\n * Beschreibung: Browser-Tabs auflisten\n * Parameter: Keine\n * Schreibgeschützt: **true**\n\n* **Browser-Tab neu**\n\n * Titel: Einen neuen Tab öffnen\n * Beschreibung: Neue Registerkarte öffnen\n * Parameter:\n * `url` (Zeichenfolge, optional): Die URL, zu der im neuen Tab navigiert werden soll. Wenn keine URL angegeben wird, ist der neue Tab leer.\n * Schreibgeschützt: **true**\n\n* **Browser-Tab-Auswahl**\n\n * Titel: Wählen Sie eine Registerkarte\n * Beschreibung: Wählen Sie eine Registerkarte nach Index aus\n * Parameter:\n * `index` (Zahl): Der Index der auszuwählenden Registerkarte\n * Schreibgeschützt: **true**\n\n* **Browser-Tab schließen**\n\n * Titel: Einen Tab schließen\n * Beschreibung: Schließen Sie einen Tab\n * Parameter:\n * `index` (Zahl, optional): Der Index der zu schließenden Registerkarte. Schließt die aktuelle Registerkarte, falls nicht angegeben.\n * Schreibgeschützt: **false**\n\n* **Browser\\_Generieren\\_Playwright\\_Test**\n\n * Titel: Generieren Sie einen Playwright-Test\n * Beschreibung: Generieren Sie einen Playwright-Test für ein bestimmtes Szenario\n * Parameter:\n * `name` (Zeichenfolge): Der Name des Tests\n * `description` (Zeichenfolge): Die Beschreibung des Tests\n * `steps` (Array): Die Schritte des Tests\n * Schreibgeschützt: **true**\n\n* **Browser-Bildschirmaufnahme**\n\n * Titel: Machen Sie einen Screenshot\n * Beschreibung: Machen Sie einen Screenshot der aktuellen Seite\n * Parameter: Keine\n * Schreibgeschützt: **true**\n\n* **Browser-Bildschirm\\_Maus bewegen**\n\n * Titel: Maus bewegen\n * Beschreibung: Maus an eine bestimmte Position bewegen\n * Parameter:\n * `element` (Zeichenfolge): Für Menschen lesbare Elementbeschreibung, die verwendet wird, um die Berechtigung zur Interaktion mit dem Element zu erhalten\n * `x` (Zahl): X-Koordinate\n * `y` (Zahl): Y-Koordinate\n * Schreibgeschützt: **true**\n\n* **Browser-Bildschirmklick**\n\n * Titel: Klick\n * Beschreibung: Klicken Sie mit der linken Maustaste\n * Parameter:\n * `element` (Zeichenfolge): Für Menschen lesbare Elementbeschreibung, die verwendet wird, um die Berechtigung zur Interaktion mit dem Element zu erhalten\n * `x` (Zahl): X-Koordinate\n * `y` (Zahl): Y-Koordinate\n * Schreibgeschützt: **false**\n\n* **Browser-Bildschirm ziehen**\n\n * Titel: Maus ziehen\n * Beschreibung: Linke Maustaste ziehen\n * Parameter:\n * `element` (Zeichenfolge): Für Menschen lesbare Elementbeschreibung, die verwendet wird, um die Berechtigung zur Interaktion mit dem Element zu erhalten\n * `startX` (Zahl): Start-X-Koordinate\n * `startY` (Zahl): Start-Y-Koordinate\n * `endX` (Zahl): End-X-Koordinate\n * `endY` (Zahl): Y-Endkoordinate\n * Schreibgeschützt: **false**\n\n* **Browser-Bildschirmtyp**\n\n * Titel: Text eingeben\n * Beschreibung: Geben Sie Text ein\n * Parameter:\n * `text` (Zeichenfolge): Text, der in das Element eingegeben werden soll\n * `submit` (Boolesch, optional): Ob eingegebener Text abgesendet werden soll (nachher die Eingabetaste drücken)\n * Schreibgeschützt: **false**\n\n* **Browser\\_Press\\_Key**\n\n * Titel: Drücken Sie eine Taste\n * Beschreibung: Drücken Sie eine Taste auf der Tastatur\n * Parameter:\n * `key` (string): Name der zu drückenden Taste oder eines zu generierenden Zeichens, wie etwa `ArrowLeft` oder `a`\n * Schreibgeschützt: **false**\n\n* **Browser\\_wait\\_for**\n\n * Titel: Warten auf\n * Beschreibung: Warten Sie, bis Text erscheint oder verschwindet oder eine bestimmte Zeit vergeht\n * Parameter:\n * `time` (Zahl, optional): Die Wartezeit in Sekunden\n * `text` (Zeichenfolge, optional): Der Text, auf den gewartet werden soll\n * `textGone` (Zeichenfolge, optional): Der Text, auf dessen Verschwinden gewartet werden soll\n * Schreibgeschützt: **true**\n\n* **Browser-Datei-Upload**\n\n * Titel: Dateien hochladen\n * Beschreibung: Eine oder mehrere Dateien hochladen\n * Parameter:\n * `paths` (Array): Die absoluten Pfade zu den hochzuladenden Dateien. Kann eine einzelne oder mehrere Dateien sein.\n * Schreibgeschützt: **false**\n\n* **Browser-Handle-Dialog**\n\n * Titel: Einen Dialog führen\n * Beschreibung: Behandeln eines Dialogs\n * Parameter:\n * `accept` (Boolesch): Ob der Dialog akzeptiert werden soll.\n * `promptText` (Zeichenfolge, optional): Der Text der Eingabeaufforderung im Falle eines Eingabeaufforderungsdialogs.\n * Schreibgeschützt: **false**","German","tag","de-DE",{"_36":48,"_40":52,"_45":53},{"_38":49,"_40":50,"_42":51},"A Model Context Protocol server that enables LLMs to interact with web pages through structured accessibility snapshots without requiring vision models or screenshots.","Playwright MCP Server","## Playwright MCP\n\nA Model Context Protocol (MCP) server that provides browser automation capabilities using [Playwright](https://playwright.dev). This server enables LLMs to interact with web pages through structured accessibility snapshots, bypassing the need for screenshots or visually-tuned models.\n\n### Key Features\n\n- **Fast and lightweight**. Uses Playwright's accessibility tree, not pixel-based input.\n- **LLM-friendly**. No vision models needed, operates purely on structured data.\n- **Deterministic tool application**. Avoids ambiguity common with screenshot-based approaches.\n\n### Requirements\n- Node.js 18 or newer\n- VS Code, Cursor, Windsurf, Claude Desktop or any other MCP client\n\n\n\n### Getting started\n\nFirst, install the Playwright MCP server with your client. A typical configuration looks like this:\n\n```js\n{\n \"mcpServers\": {\n \"playwright\": {\n \"command\": \"npx\",\n \"args\": [\n \"@playwright/mcp@latest\"\n ]\n }\n }\n}\n```\n\n[ $\"Install$ ](https://insiders.vscode.dev/redirect?url=vscode%3Amcp%2Finstall%3F%257B%2522name%2522%253A%2522playwright%2522%252C%2522command%2522%253A%2522npx%2522%252C%2522args%2522%253A%255B%2522%2540playwright%252Fmcp%2540latest%2522%255D%257D) [ $\"Install$ ](https://insiders.vscode.dev/redirect?url=vscode-insiders%3Amcp%2Finstall%3F%257B%2522name%2522%253A%2522playwright%2522%252C%2522command%2522%253A%2522npx%2522%252C%2522args%2522%253A%255B%2522%2540playwright%252Fmcp%2540latest%2522%255D%257D)\n\n\n

Install in VS Code

\n\nYou can also install the Playwright MCP server using the VS Code CLI:\n\n```bash\n# For VS Code\ncode --add-mcp '{\"name\":\"playwright\",\"command\":\"npx\",\"args\":[\"@playwright/mcp@latest\"]}'\n```\n\nAfter installation, the Playwright MCP server will be available for use with your GitHub Copilot agent in VS Code.\n

\n\n

Install in Cursor

\n\nGo to `Cursor Settings` -> `MCP` -> `Add new MCP Server`. Name to your liking, use `command` type with the command `npx @playwright/mcp`. You can also verify config or add command like arguments via clicking `Edit`.\n\n```js\n{\n \"mcpServers\": {\n \"playwright\": {\n \"command\": \"npx\",\n \"args\": [\n \"@playwright/mcp@latest\"\n ]\n }\n }\n}\n```\n

\n\n

Install in Windsurf

\n\nFollow Windsuff MCP [documentation](https://docs.windsurf.com/windsurf/cascade/mcp). Use following configuration:\n\n```js\n{\n \"mcpServers\": {\n \"playwright\": {\n \"command\": \"npx\",\n \"args\": [\n \"@playwright/mcp@latest\"\n ]\n }\n }\n}\n```\n

\n\n

Install in Claude Desktop

\n\nFollow the MCP install [guide](https://modelcontextprotocol.io/quickstart/user), use following configuration:\n\n```js\n{\n \"mcpServers\": {\n \"playwright\": {\n \"command\": \"npx\",\n \"args\": [\n \"@playwright/mcp@latest\"\n ]\n }\n }\n}\n```\n

\n\n### Configuration\n\nPlaywright MCP server supports following arguments. They can be provided in the JSON configuration above, as a part of the `\"args\"` list:\n\n\n\n```\n> npx @playwright/mcp@latest --help\n --allowed-origins semicolon-separated list of origins to allow the\n browser to request. Default is to allow all.\n --blocked-origins semicolon-separated list of origins to block the\n browser from requesting. Blocklist is evaluated\n before allowlist. If used without the allowlist,\n requests not matching the blocklist are still\n allowed.\n --block-service-workers block service workers\n --browser browser or chrome channel to use, possible\n values: chrome, firefox, webkit, msedge.\n --caps comma-separated list of capabilities to enable,\n possible values: tabs, pdf, history, wait, files,\n install. Default is all.\n --cdp-endpoint CDP endpoint to connect to.\n --config path to the configuration file.\n --device device to emulate, for example: \"iPhone 15\"\n --executable-path path to the browser executable.\n --headless run browser in headless mode, headed by default\n --host host to bind server to. Default is localhost. Use\n 0.0.0.0 to bind to all interfaces.\n --ignore-https-errors ignore https errors\n --isolated keep the browser profile in memory, do not save\n it to disk.\n --no-image-responses do not send image responses to the client.\n --no-sandbox disable the sandbox for all process types that\n are normally sandboxed.\n --output-dir path to the directory for output files.\n --port port to listen on for SSE transport.\n --proxy-bypass comma-separated domains to bypass proxy, for\n example \".com,chromium.org,.domain.com\"\n --proxy-server specify proxy server, for example\n \"http://myproxy:3128\" or \"socks5://myproxy:8080\"\n --save-trace Whether to save the Playwright Trace of the\n session into the output directory.\n --storage-state path to the storage state file for isolated\n sessions.\n --user-agent specify user agent string\n --user-data-dir path to the user data directory. If not\n specified, a temporary directory will be created.\n --viewport-size specify browser viewport size in pixels, for\n example \"1280, 720\"\n --vision Run server that uses screenshots (Aria snapshots\n are used by default)\n```\n\n\n\n### User profile\n\nYou can run Playwright MCP with persistent profile like a regular browser (default), or in the isolated contexts for the testing sessions.\n\n**Persistent profile**\n\nAll the logged in information will be stored in the persistent profile, you can delete it between sessions if you'd like to clear the offline state.\nPersistent profile is located at the following locations and you can override it with the `--user-data-dir` argument.\n\n```bash\n# Windows\n%USERPROFILE%\\AppData\\Local\\ms-playwright\\mcp-{channel}-profile\n\n# macOS\n- ~/Library/Caches/ms-playwright/mcp-{channel}-profile\n\n# Linux\n- ~/.cache/ms-playwright/mcp-{channel}-profile\n```\n\n**Isolated**\n\nIn the isolated mode, each session is started in the isolated profile. Every time you ask MCP to close the browser,\nthe session is closed and all the storage state for this session is lost. You can provide initial storage state\nto the browser via the config's `contextOptions` or via the `--storage-state` argument. Learn more about the storage\nstate [here](https://playwright.dev/docs/auth).\n\n```js\n{\n \"mcpServers\": {\n \"playwright\": {\n \"command\": \"npx\",\n \"args\": [\n \"@playwright/mcp@latest\",\n \"--isolated\",\n \"--storage-state={path/to/storage.json}\n ]\n }\n }\n}\n```\n\n### Configuration file\n\nThe Playwright MCP server can be configured using a JSON configuration file. You can specify the configuration file\nusing the `--config` command line option:\n\n```bash\nnpx @playwright/mcp@latest --config path/to/config.json\n```\n\n

Configuration file schema

\n\n```typescript\n{\n // Browser configuration\n browser?: {\n // Browser type to use (chromium, firefox, or webkit)\n browserName?: 'chromium' | 'firefox' | 'webkit';\n\n // Keep the browser profile in memory, do not save it to disk.\n isolated?: boolean;\n\n // Path to user data directory for browser profile persistence\n userDataDir?: string;\n\n // Browser launch options (see Playwright docs)\n // @see https://playwright.dev/docs/api/class-browsertype#browser-type-launch\n launchOptions?: {\n channel?: string; // Browser channel (e.g. 'chrome')\n headless?: boolean; // Run in headless mode\n executablePath?: string; // Path to browser executable\n // ... other Playwright launch options\n };\n\n // Browser context options\n // @see https://playwright.dev/docs/api/class-browser#browser-new-context\n contextOptions?: {\n viewport?: { width: number, height: number };\n // ... other Playwright context options\n };\n\n // CDP endpoint for connecting to existing browser\n cdpEndpoint?: string;\n\n // Remote Playwright server endpoint\n remoteEndpoint?: string;\n },\n\n // Server configuration\n server?: {\n port?: number; // Port to listen on\n host?: string; // Host to bind to (default: localhost)\n },\n\n // List of enabled capabilities\n capabilities?: Array<\n 'core' | // Core browser automation\n 'tabs' | // Tab management\n 'pdf' | // PDF generation\n 'history' | // Browser history\n 'wait' | // Wait utilities\n 'files' | // File handling\n 'install' | // Browser installation\n 'testing' // Testing\n >;\n\n // Enable vision mode (screenshots instead of accessibility snapshots)\n vision?: boolean;\n\n // Directory for output files\n outputDir?: string;\n\n // Network configuration\n network?: {\n // List of origins to allow the browser to request. Default is to allow all. Origins matching both `allowedOrigins` and `blockedOrigins` will be blocked.\n allowedOrigins?: string[];\n\n // List of origins to block the browser to request. Origins matching both `allowedOrigins` and `blockedOrigins` will be blocked.\n blockedOrigins?: string[];\n };\n \n /**\n * Do not send image responses to the client.\n */\n noImageResponses?: boolean;\n}\n```\n

\n\n### Standalone MCP server\n\nWhen running headed browser on system w/o display or from worker processes of the IDEs,\nrun the MCP server from environment with the DISPLAY and pass the `--port` flag to enable SSE transport.\n\n```bash\nnpx @playwright/mcp@latest --port 8931\n```\n\nAnd then in MCP client config, set the `url` to the SSE endpoint:\n\n```js\n{\n \"mcpServers\": {\n \"playwright\": {\n \"url\": \"http://localhost:8931/sse\"\n }\n }\n}\n```\n\n

Docker

\n\n**NOTE:** The Docker implementation only supports headless chromium at the moment.\n\n```js\n{\n \"mcpServers\": {\n \"playwright\": {\n \"command\": \"docker\",\n \"args\": [\"run\", \"-i\", \"--rm\", \"--init\", \"--pull=always\", \"mcr.microsoft.com/playwright/mcp\"]\n }\n }\n}\n```\n\nYou can build the Docker image yourself.\n\n```\ndocker build -t mcr.microsoft.com/playwright/mcp .\n```\n

\n\n

Programmatic usage

\n\n```js\nimport http from 'http';\n\nimport { createConnection } from '@playwright/mcp';\nimport { SSEServerTransport } from '@modelcontextprotocol/sdk/server/sse.js';\n\nhttp.createServer(async (req, res) => {\n // ...\n\n // Creates a headless Playwright MCP server with SSE transport\n const connection = await createConnection({ browser: { launchOptions: { headless: true } } });\n const transport = new SSEServerTransport('/messages', res);\n await connection.connect(transport);\n\n // ...\n});\n```\n

\n\n### Tools\n\nThe tools are available in two modes:\n\n1. **Snapshot Mode** (default): Uses accessibility snapshots for better performance and reliability\n2. **Vision Mode**: Uses screenshots for visual-based interactions\n\nTo use Vision Mode, add the `--vision` flag when starting the server:\n\n```js\n{\n \"mcpServers\": {\n \"playwright\": {\n \"command\": \"npx\",\n \"args\": [\n \"@playwright/mcp@latest\",\n \"--vision\"\n ]\n }\n }\n}\n```\n\nVision Mode works best with the computer use models that are able to interact with elements using\nX Y coordinate space, based on the provided screenshot.\n\n\n\n

Interactions

\n\n\n\n- **browser_snapshot**\n - Title: Page snapshot\n - Description: Capture accessibility snapshot of the current page, this is better than screenshot\n - Parameters: None\n - Read-only: **true**\n\n\n\n- **browser_click**\n - Title: Click\n - Description: Perform click on a web page\n - Parameters:\n - `element` (string): Human-readable element description used to obtain permission to interact with the element\n - `ref` (string): Exact target element reference from the page snapshot\n - Read-only: **false**\n\n\n\n- **browser_drag**\n - Title: Drag mouse\n - Description: Perform drag and drop between two elements\n - Parameters:\n - `startElement` (string): Human-readable source element description used to obtain the permission to interact with the element\n - `startRef` (string): Exact source element reference from the page snapshot\n - `endElement` (string): Human-readable target element description used to obtain the permission to interact with the element\n - `endRef` (string): Exact target element reference from the page snapshot\n - Read-only: **false**\n\n\n\n- **browser_hover**\n - Title: Hover mouse\n - Description: Hover over element on page\n - Parameters:\n - `element` (string): Human-readable element description used to obtain permission to interact with the element\n - `ref` (string): Exact target element reference from the page snapshot\n - Read-only: **true**\n\n\n\n- **browser_type**\n - Title: Type text\n - Description: Type text into editable element\n - Parameters:\n - `element` (string): Human-readable element description used to obtain permission to interact with the element\n - `ref` (string): Exact target element reference from the page snapshot\n - `text` (string): Text to type into the element\n - `submit` (boolean, optional): Whether to submit entered text (press Enter after)\n - `slowly` (boolean, optional): Whether to type one character at a time. Useful for triggering key handlers in the page. By default entire text is filled in at once.\n - Read-only: **false**\n\n\n\n- **browser_select_option**\n - Title: Select option\n - Description: Select an option in a dropdown\n - Parameters:\n - `element` (string): Human-readable element description used to obtain permission to interact with the element\n - `ref` (string): Exact target element reference from the page snapshot\n - `values` (array): Array of values to select in the dropdown. This can be a single value or multiple values.\n - Read-only: **false**\n\n\n\n- **browser_press_key**\n - Title: Press a key\n - Description: Press a key on the keyboard\n - Parameters:\n - `key` (string): Name of the key to press or a character to generate, such as `ArrowLeft` or `a`\n - Read-only: **false**\n\n\n\n- **browser_wait_for**\n - Title: Wait for\n - Description: Wait for text to appear or disappear or a specified time to pass\n - Parameters:\n - `time` (number, optional): The time to wait in seconds\n - `text` (string, optional): The text to wait for\n - `textGone` (string, optional): The text to wait for to disappear\n - Read-only: **true**\n\n\n\n- **browser_file_upload**\n - Title: Upload files\n - Description: Upload one or multiple files\n - Parameters:\n - `paths` (array): The absolute paths to the files to upload. Can be a single file or multiple files.\n - Read-only: **false**\n\n\n\n- **browser_handle_dialog**\n - Title: Handle a dialog\n - Description: Handle a dialog\n - Parameters:\n - `accept` (boolean): Whether to accept the dialog.\n - `promptText` (string, optional): The text of the prompt in case of a prompt dialog.\n - Read-only: **false**\n\n

\n\n

Navigation

\n\n\n\n- **browser_navigate**\n - Title: Navigate to a URL\n - Description: Navigate to a URL\n - Parameters:\n - `url` (string): The URL to navigate to\n - Read-only: **false**\n\n\n\n- **browser_navigate_back**\n - Title: Go back\n - Description: Go back to the previous page\n - Parameters: None\n - Read-only: **true**\n\n\n\n- **browser_navigate_forward**\n - Title: Go forward\n - Description: Go forward to the next page\n - Parameters: None\n - Read-only: **true**\n\n

\n\n

Resources

\n\n\n\n- **browser_take_screenshot**\n - Title: Take a screenshot\n - Description: Take a screenshot of the current page. You can't perform actions based on the screenshot, use browser_snapshot for actions.\n - Parameters:\n - `raw` (boolean, optional): Whether to return without compression (in PNG format). Default is false, which returns a JPEG image.\n - `filename` (string, optional): File name to save the screenshot to. Defaults to `page-{timestamp}.{png|jpeg}` if not specified.\n - `element` (string, optional): Human-readable element description used to obtain permission to screenshot the element. If not provided, the screenshot will be taken of viewport. If element is provided, ref must be provided too.\n - `ref` (string, optional): Exact target element reference from the page snapshot. If not provided, the screenshot will be taken of viewport. If ref is provided, element must be provided too.\n - Read-only: **true**\n\n\n\n- **browser_pdf_save**\n - Title: Save as PDF\n - Description: Save page as PDF\n - Parameters:\n - `filename` (string, optional): File name to save the pdf to. Defaults to `page-{timestamp}.pdf` if not specified.\n - Read-only: **true**\n\n\n\n- **browser_network_requests**\n - Title: List network requests\n - Description: Returns all network requests since loading the page\n - Parameters: None\n - Read-only: **true**\n\n\n\n- **browser_console_messages**\n - Title: Get console messages\n - Description: Returns all console messages\n - Parameters: None\n - Read-only: **true**\n\n

\n\n

Utilities

\n\n\n\n- **browser_install**\n - Title: Install the browser specified in the config\n - Description: Install the browser specified in the config. Call this if you get an error about the browser not being installed.\n - Parameters: None\n - Read-only: **false**\n\n\n\n- **browser_close**\n - Title: Close browser\n - Description: Close the page\n - Parameters: None\n - Read-only: **true**\n\n\n\n- **browser_resize**\n - Title: Resize browser window\n - Description: Resize the browser window\n - Parameters:\n - `width` (number): Width of the browser window\n - `height` (number): Height of the browser window\n - Read-only: **true**\n\n

\n\n

Tabs

\n\n\n\n- **browser_tab_list**\n - Title: List tabs\n - Description: List browser tabs\n - Parameters: None\n - Read-only: **true**\n\n\n\n- **browser_tab_new**\n - Title: Open a new tab\n - Description: Open a new tab\n - Parameters:\n - `url` (string, optional): The URL to navigate to in the new tab. If not provided, the new tab will be blank.\n - Read-only: **true**\n\n\n\n- **browser_tab_select**\n - Title: Select a tab\n - Description: Select a tab by index\n - Parameters:\n - `index` (number): The index of the tab to select\n - Read-only: **true**\n\n\n\n- **browser_tab_close**\n - Title: Close a tab\n - Description: Close a tab\n - Parameters:\n - `index` (number, optional): The index of the tab to close. Closes current tab if not provided.\n - Read-only: **false**\n\n

\n\n

Testing

\n\n\n\n- **browser_generate_playwright_test**\n - Title: Generate a Playwright test\n - Description: Generate a Playwright test for given scenario\n - Parameters:\n - `name` (string): The name of the test\n - `description` (string): The description of the test\n - `steps` (array): The steps of the test\n - Read-only: **true**\n\n

\n\n

Vision mode

\n\n\n\n- **browser_screen_capture**\n - Title: Take a screenshot\n - Description: Take a screenshot of the current page\n - Parameters: None\n - Read-only: **true**\n\n\n\n- **browser_screen_move_mouse**\n - Title: Move mouse\n - Description: Move mouse to a given position\n - Parameters:\n - `element` (string): Human-readable element description used to obtain permission to interact with the element\n - `x` (number): X coordinate\n - `y` (number): Y coordinate\n - Read-only: **true**\n\n\n\n- **browser_screen_click**\n - Title: Click\n - Description: Click left mouse button\n - Parameters:\n - `element` (string): Human-readable element description used to obtain permission to interact with the element\n - `x` (number): X coordinate\n - `y` (number): Y coordinate\n - Read-only: **false**\n\n\n\n- **browser_screen_drag**\n - Title: Drag mouse\n - Description: Drag left mouse button\n - Parameters:\n - `element` (string): Human-readable element description used to obtain permission to interact with the element\n - `startX` (number): Start X coordinate\n - `startY` (number): Start Y coordinate\n - `endX` (number): End X coordinate\n - `endY` (number): End Y coordinate\n - Read-only: **false**\n\n\n\n- **browser_screen_type**\n - Title: Type text\n - Description: Type text\n - Parameters:\n - `text` (string): Text to type into the element\n - `submit` (boolean, optional): Whether to submit entered text (press Enter after)\n - Read-only: **false**\n\n\n\n- **browser_press_key**\n - Title: Press a key\n - Description: Press a key on the keyboard\n - Parameters:\n - `key` (string): Name of the key to press or a character to generate, such as `ArrowLeft` or `a`\n - Read-only: **false**\n\n\n\n- **browser_wait_for**\n - Title: Wait for\n - Description: Wait for text to appear or disappear or a specified time to pass\n - Parameters:\n - `time` (number, optional): The time to wait in seconds\n - `text` (string, optional): The text to wait for\n - `textGone` (string, optional): The text to wait for to disappear\n - Read-only: **true**\n\n\n\n- **browser_file_upload**\n - Title: Upload files\n - Description: Upload one or multiple files\n - Parameters:\n - `paths` (array): The absolute paths to the files to upload. Can be a single file or multiple files.\n - Read-only: **false**\n\n\n\n- **browser_handle_dialog**\n - Title: Handle a dialog\n - Description: Handle a dialog\n - Parameters:\n - `accept` (boolean): Whether to accept the dialog.\n - `promptText` (string, optional): The text of the prompt in case of a prompt dialog.\n - Read-only: **false**\n\n

\n\n\n\n","English","en-US",{"_36":55,"_40":59,"_45":60},{"_38":56,"_40":57,"_42":58},"Un servidor de protocolo de contexto de modelo que permite a los LLM interactuar con páginas web a través de instantáneas de accesibilidad estructuradas sin necesidad de modelos de visión o capturas de pantalla.","Servidor MCP del dramaturgo","## Dramaturgo MCP\n\nUn servidor de Protocolo de Contexto de Modelo (MCP) que proporciona funciones de automatización del navegador mediante [Playwright](https://playwright.dev) . Este servidor permite a los LLM interactuar con páginas web mediante instantáneas de accesibilidad estructuradas, eliminando la necesidad de capturas de pantalla o modelos visualmente optimizados.\n\n### Características principales\n\n* **Rápido y ligero** . Utiliza el árbol de accesibilidad de Playwright, no la entrada basada en píxeles.\n* **Compatible con LLM** . No requiere modelos de visión; opera exclusivamente con datos estructurados.\n* **Aplicación de herramientas deterministas** . Evita la ambigüedad común en los enfoques basados en capturas de pantalla.\n\n### Requisitos\n\n* Node.js 18 o más reciente\n* VS Code, Cursor, Windsurf, Claude Desktop o cualquier otro cliente MCP\n\n### Empezando\n\nPrimero, instala el servidor Playwright MCP con tu cliente. Una configuración típica es la siguiente:\n\n```js\n{\n \"mcpServers\": {\n \"playwright\": {\n \"command\": \"npx\",\n \"args\": [\n \"@playwright/mcp@latest\"\n ]\n }\n }\n}\n```\n\nTambién puedes instalar el servidor Playwright MCP usando la CLI de VS Code:\n\n```bash\n# For VS Code\ncode --add-mcp '{\"name\":\"playwright\",\"command\":\"npx\",\"args\":[\"@playwright/mcp@latest\"]}'\n```\n\nDespués de la instalación, el servidor Playwright MCP estará disponible para su uso con su agente GitHub Copilot en VS Code.\n\nVaya a `Cursor Settings` -> `MCP` -> `Add new MCP Server` . Asigne un nombre a su gusto y `command` el comando `npx @playwright/mcp` . También puede verificar la configuración o agregar argumentos similares a los de un comando haciendo clic en `Edit` .\n\n```js\n{\n \"mcpServers\": {\n \"playwright\": {\n \"command\": \"npx\",\n \"args\": [\n \"@playwright/mcp@latest\"\n ]\n }\n }\n}\n```\n\nConsulte [la documentación de](https://docs.windsurf.com/windsurf/cascade/mcp) Windsuff MCP. Utilice la siguiente configuración:\n\n```js\n{\n \"mcpServers\": {\n \"playwright\": {\n \"command\": \"npx\",\n \"args\": [\n \"@playwright/mcp@latest\"\n ]\n }\n }\n}\n```\n\nSiga la [guía](https://modelcontextprotocol.io/quickstart/user) de instalación de MCP, utilice la siguiente configuración:\n\n```js\n{\n \"mcpServers\": {\n \"playwright\": {\n \"command\": \"npx\",\n \"args\": [\n \"@playwright/mcp@latest\"\n ]\n }\n }\n}\n```\n\n### Configuración\n\nEl servidor Playwright MCP admite los siguientes argumentos. Se pueden proporcionar en la configuración JSON anterior, como parte de la lista `\"args\"` :\n\n```\n> npx @playwright/mcp@latest --help\n --allowed-origins semicolon-separated list of origins to allow the\n browser to request. Default is to allow all.\n --blocked-origins semicolon-separated list of origins to block the\n browser from requesting. Blocklist is evaluated\n before allowlist. If used without the allowlist,\n requests not matching the blocklist are still\n allowed.\n --block-service-workers block service workers\n --browser browser or chrome channel to use, possible\n values: chrome, firefox, webkit, msedge.\n --caps comma-separated list of capabilities to enable,\n possible values: tabs, pdf, history, wait, files,\n install. Default is all.\n --cdp-endpoint CDP endpoint to connect to.\n --config path to the configuration file.\n --device device to emulate, for example: \"iPhone 15\"\n --executable-path path to the browser executable.\n --headless run browser in headless mode, headed by default\n --host host to bind server to. Default is localhost. Use\n 0.0.0.0 to bind to all interfaces.\n --ignore-https-errors ignore https errors\n --isolated keep the browser profile in memory, do not save\n it to disk.\n --no-image-responses do not send image responses to the client.\n --no-sandbox disable the sandbox for all process types that\n are normally sandboxed.\n --output-dir path to the directory for output files.\n --port port to listen on for SSE transport.\n --proxy-bypass comma-separated domains to bypass proxy, for\n example \".com,chromium.org,.domain.com\"\n --proxy-server specify proxy server, for example\n \"http://myproxy:3128\" or \"socks5://myproxy:8080\"\n --save-trace Whether to save the Playwright Trace of the\n session into the output directory.\n --storage-state path to the storage state file for isolated\n sessions.\n --user-agent specify user agent string\n --user-data-dir path to the user data directory. If not\n specified, a temporary directory will be created.\n --viewport-size specify browser viewport size in pixels, for\n example \"1280, 720\"\n --vision Run server that uses screenshots (Aria snapshots\n are used by default)\n```\n\n### Perfil de usuario\n\nPuede ejecutar Playwright MCP con un perfil persistente como un navegador normal (predeterminado) o en contextos aislados para las sesiones de prueba.\n\n**Perfil persistente**\n\nToda la información de inicio de sesión se almacenará en el perfil persistente. Puede eliminarlo entre sesiones si desea desactivar el estado sin conexión. El perfil persistente se encuentra en las siguientes ubicaciones y puede sobrescribirlo con el argumento `--user-data-dir` .\n\n```bash\n# Windows\n%USERPROFILE%\\AppData\\Local\\ms-playwright\\mcp-{channel}-profile\n\n# macOS\n- ~/Library/Caches/ms-playwright/mcp-{channel}-profile\n\n# Linux\n- ~/.cache/ms-playwright/mcp-{channel}-profile\n```\n\n**Aislado**\n\nEn el modo aislado, cada sesión se inicia en el perfil aislado. Cada vez que se solicita a MCP que cierre el navegador, la sesión se cierra y se pierde todo el estado de almacenamiento de esta sesión. Puede proporcionar el estado de almacenamiento inicial al navegador mediante las `contextOptions` de la configuración o el argumento `--storage-state` . Obtenga más información sobre el estado de almacenamiento [aquí](https://playwright.dev/docs/auth) .\n\n```js\n{\n \"mcpServers\": {\n \"playwright\": {\n \"command\": \"npx\",\n \"args\": [\n \"@playwright/mcp@latest\",\n \"--isolated\",\n \"--storage-state={path/to/storage.json}\n ]\n }\n }\n}\n```\n\n### Archivo de configuración\n\nEl servidor Playwright MCP se puede configurar mediante un archivo de configuración JSON. Puede especificar el archivo de configuración con la opción de línea de comandos `--config` :\n\n```bash\nnpx @playwright/mcp@latest --config path/to/config.json\n```\n\n```typescript\n{\n // Browser configuration\n browser?: {\n // Browser type to use (chromium, firefox, or webkit)\n browserName?: 'chromium' | 'firefox' | 'webkit';\n\n // Keep the browser profile in memory, do not save it to disk.\n isolated?: boolean;\n\n // Path to user data directory for browser profile persistence\n userDataDir?: string;\n\n // Browser launch options (see Playwright docs)\n // @see https://playwright.dev/docs/api/class-browsertype#browser-type-launch\n launchOptions?: {\n channel?: string; // Browser channel (e.g. 'chrome')\n headless?: boolean; // Run in headless mode\n executablePath?: string; // Path to browser executable\n // ... other Playwright launch options\n };\n\n // Browser context options\n // @see https://playwright.dev/docs/api/class-browser#browser-new-context\n contextOptions?: {\n viewport?: { width: number, height: number };\n // ... other Playwright context options\n };\n\n // CDP endpoint for connecting to existing browser\n cdpEndpoint?: string;\n\n // Remote Playwright server endpoint\n remoteEndpoint?: string;\n },\n\n // Server configuration\n server?: {\n port?: number; // Port to listen on\n host?: string; // Host to bind to (default: localhost)\n },\n\n // List of enabled capabilities\n capabilities?: Array<\n 'core' | // Core browser automation\n 'tabs' | // Tab management\n 'pdf' | // PDF generation\n 'history' | // Browser history\n 'wait' | // Wait utilities\n 'files' | // File handling\n 'install' | // Browser installation\n 'testing' // Testing\n >;\n\n // Enable vision mode (screenshots instead of accessibility snapshots)\n vision?: boolean;\n\n // Directory for output files\n outputDir?: string;\n\n // Network configuration\n network?: {\n // List of origins to allow the browser to request. Default is to allow all. Origins matching both `allowedOrigins` and `blockedOrigins` will be blocked.\n allowedOrigins?: string[];\n\n // List of origins to block the browser to request. Origins matching both `allowedOrigins` and `blockedOrigins` will be blocked.\n blockedOrigins?: string[];\n };\n \n /**\n * Do not send image responses to the client.\n */\n noImageResponses?: boolean;\n}\n```\n\n### Servidor MCP independiente\n\nAl ejecutar el navegador en un sistema sin pantalla o desde procesos de trabajo de los IDE, ejecute el servidor MCP desde el entorno con PANTALLA y pase el indicador `--port` para habilitar el transporte SSE.\n\n```bash\nnpx @playwright/mcp@latest --port 8931\n```\n\nY luego, en la configuración del cliente MCP, configure la `url` del punto final SSE:\n\n```js\n{\n \"mcpServers\": {\n \"playwright\": {\n \"url\": \"http://localhost:8931/sse\"\n }\n }\n}\n```\n\n**NOTA:** La implementación de Docker solo admite Chromium sin interfaz gráfica por el momento.\n\n```js\n{\n \"mcpServers\": {\n \"playwright\": {\n \"command\": \"docker\",\n \"args\": [\"run\", \"-i\", \"--rm\", \"--init\", \"--pull=always\", \"mcr.microsoft.com/playwright/mcp\"]\n }\n }\n}\n```\n\nPuedes crear la imagen de Docker tú mismo.\n\n```\ndocker build -t mcr.microsoft.com/playwright/mcp .\n```\n\n```js\nimport http from 'http';\n\nimport { createConnection } from '@playwright/mcp';\nimport { SSEServerTransport } from '@modelcontextprotocol/sdk/server/sse.js';\n\nhttp.createServer(async (req, res) => {\n // ...\n\n // Creates a headless Playwright MCP server with SSE transport\n const connection = await createConnection({ browser: { launchOptions: { headless: true } } });\n const transport = new SSEServerTransport('/messages', res);\n await connection.connect(transport);\n\n // ...\n});\n```\n\n### Herramientas\n\nLas herramientas están disponibles en dos modos:\n\n1. **Modo instantánea** (predeterminado): utiliza instantáneas de accesibilidad para un mejor rendimiento y confiabilidad\n2. **Modo visión** : utiliza capturas de pantalla para interacciones visuales.\n\nPara utilizar el modo Visión, agregue el indicador `--vision` al iniciar el servidor:\n\n```js\n{\n \"mcpServers\": {\n \"playwright\": {\n \"command\": \"npx\",\n \"args\": [\n \"@playwright/mcp@latest\",\n \"--vision\"\n ]\n }\n }\n}\n```\n\nEl modo Visión funciona mejor con los modelos de uso de computadora que pueden interactuar con elementos utilizando el espacio de coordenadas XY, según la captura de pantalla proporcionada.\n\n* **instantánea del navegador**\n\n * Título: Instantánea de la página\n * Descripción: Captura una instantánea de accesibilidad de la página actual, esto es mejor que una captura de pantalla.\n * Parámetros: Ninguno\n * Solo lectura: **verdadero**\n\n* **clic del navegador**\n\n * Título: Haga clic\n * Descripción: Realizar clic en una página web\n * Parámetros:\n * `element` (cadena): Descripción del elemento legible por humanos que se utiliza para obtener permiso para interactuar con el elemento\n * `ref` (cadena): referencia exacta del elemento de destino de la instantánea de la página\n * Solo lectura: **falso**\n\n* **arrastrar\\_del\\_navegador**\n\n * Título: Arrastre el ratón\n * Descripción: Realizar arrastrar y soltar entre dos elementos\n * Parámetros:\n * `startElement` (cadena): descripción del elemento fuente legible por humanos que se utiliza para obtener el permiso para interactuar con el elemento\n * `startRef` (cadena): referencia exacta del elemento fuente de la instantánea de la página\n * `endElement` (cadena): descripción del elemento de destino legible por humanos que se utiliza para obtener el permiso para interactuar con el elemento\n * `endRef` (cadena): referencia exacta del elemento de destino de la instantánea de la página\n * Solo lectura: **falso**\n\n* **pasar el cursor por el navegador**\n\n * Título: Pasar el ratón\n * Descripción: Pase el cursor sobre el elemento en la página\n * Parámetros:\n * `element` (cadena): Descripción del elemento legible por humanos que se utiliza para obtener permiso para interactuar con el elemento\n * `ref` (cadena): referencia exacta del elemento de destino de la instantánea de la página\n * Solo lectura: **verdadero**\n\n* **tipo de navegador**\n\n * Título: Escriba el texto\n * Descripción: Escriba texto en un elemento editable\n * Parámetros:\n * `element` (cadena): Descripción del elemento legible por humanos que se utiliza para obtener permiso para interactuar con el elemento\n * `ref` (cadena): referencia exacta del elemento de destino de la instantánea de la página\n * `text` (cadena): Texto a escribir en el elemento\n * `submit` (booleano, opcional): si se debe enviar el texto ingresado (presione Enter después)\n * `slowly` (booleano, opcional): Indica si se escribe un carácter a la vez. Útil para activar los controladores de teclas en la página. Por defecto, todo el texto se completa de una vez.\n * Solo lectura: **falso**\n\n* **opción de selección del navegador**\n\n * Título: Seleccionar opción\n * Descripción: Seleccione una opción en un menú desplegable\n * Parámetros:\n * `element` (cadena): Descripción del elemento legible por humanos que se utiliza para obtener permiso para interactuar con el elemento\n * `ref` (cadena): referencia exacta del elemento de destino de la instantánea de la página\n * `values` (matriz): Matriz de valores para seleccionar en el menú desplegable. Puede ser un solo valor o varios.\n * Solo lectura: **falso**\n\n* **tecla\\_pulsar\\_del\\_navegador**\n\n * Título: Presiona una tecla\n * Descripción: Presione una tecla del teclado\n * Parámetros:\n * `key` (cadena): Nombre de la tecla a presionar o un carácter a generar, como `ArrowLeft` o `a`\n * Solo lectura: **falso**\n\n* **espera del navegador**\n\n * Título: Espera\n * Descripción: Esperar a que aparezca o desaparezca el texto o que transcurra un tiempo específico\n * Parámetros:\n * `time` (número, opcional): el tiempo de espera en segundos\n * `text` (cadena, opcional): el texto que se debe esperar\n * `textGone` (cadena, opcional): el texto que se debe esperar a que desaparezca\n * Solo lectura: **verdadero**\n\n* **carga de archivos del navegador**\n\n * Título: Subir archivos\n * Descripción: Subir uno o varios archivos\n * Parámetros:\n * `paths` (matriz): Las rutas absolutas de los archivos que se van a cargar. Pueden ser uno o varios archivos.\n * Solo lectura: **falso**\n\n* **diálogo del identificador del navegador**\n\n * Título: Manejar un diálogo\n * Descripción: Manejar un diálogo\n * Parámetros:\n * `accept` (booleano): si se debe aceptar el diálogo.\n * `promptText` (cadena, opcional): el texto del mensaje en caso de un cuadro de diálogo de mensaje.\n * Solo lectura: **falso**\n\n* **navegador\\_navegar**\n\n * Título: Navegar a una URL\n * Descripción: Navegar a una URL\n * Parámetros:\n * `url` (cadena): La URL a la que navegar\n * Solo lectura: **falso**\n\n* **navegador\\_navegar\\_atrás**\n\n * Título: Regresar\n * Descripción: Volver a la página anterior\n * Parámetros: Ninguno\n * Solo lectura: **verdadero**\n\n* **navegador\\_navegar\\_hacia\\_adelante**\n\n * Título: Avanzar\n * Descripción: Avanzar a la página siguiente\n * Parámetros: Ninguno\n * Solo lectura: **verdadero**\n\n* **captura de pantalla del navegador**\n\n * Título: Tomar una captura de pantalla\n * Descripción: Toma una captura de pantalla de la página actual. No se pueden realizar acciones basándose en la captura de pantalla; para ello, usa \\`browser\\_snapshot\\`.\n * Parámetros:\n * `raw` (booleano, opcional): Indica si se retorna sin compresión (en formato PNG). El valor predeterminado es falso, lo que devuelve una imagen JPEG.\n * `filename` (cadena, opcional): Nombre del archivo donde se guardará la captura de pantalla. El valor predeterminado es `page-{timestamp}.{png|jpeg}` si no se especifica.\n * `element` (cadena, opcional): Descripción legible del elemento que se utiliza para obtener permiso para capturar la pantalla del elemento. Si no se proporciona, la captura se realizará desde la ventana gráfica. Si se proporciona el elemento, también se debe proporcionar la referencia.\n * `ref` (cadena, opcional): Referencia exacta del elemento de destino de la captura de pantalla de la página. Si no se proporciona, se tomará una captura de pantalla de la ventana gráfica. Si se proporciona ref, también se debe proporcionar element.\n * Solo lectura: **verdadero**\n\n* **guardar\\_pdf\\_del\\_navegador**\n\n * Título: Guardar como PDF\n * Descripción: Guardar página como PDF\n * Parámetros:\n * `filename` (cadena, opcional): nombre del archivo donde se guardará el PDF. El valor predeterminado es `page-{timestamp}.pdf` si no se especifica.\n * Solo lectura: **verdadero**\n\n* **solicitudes de red del navegador**\n\n * Título: Lista de solicitudes de red\n * Descripción: Devuelve todas las solicitudes de red desde que se cargó la página.\n * Parámetros: Ninguno\n * Solo lectura: **verdadero**\n\n* **mensajes de la consola del navegador**\n\n * Título: Obtener mensajes de la consola\n * Descripción: Devuelve todos los mensajes de la consola.\n * Parámetros: Ninguno\n * Solo lectura: **verdadero**\n\n* **instalación del navegador**\n\n * Título: Instalar el navegador especificado en la configuración\n * Descripción: Instala el navegador especificado en la configuración. Llámalo si recibes un error indicando que el navegador no está instalado.\n * Parámetros: Ninguno\n * Solo lectura: **falso**\n\n* **cerrar navegador**\n\n * Título: Cerrar navegador\n * Descripción: Cerrar la página\n * Parámetros: Ninguno\n * Solo lectura: **verdadero**\n\n* **cambio de tamaño del navegador**\n\n * Título: Cambiar el tamaño de la ventana del navegador\n * Descripción: Cambiar el tamaño de la ventana del navegador\n * Parámetros:\n * `width` (número): Ancho de la ventana del navegador\n * `height` (número): Altura de la ventana del navegador\n * Solo lectura: **verdadero**\n\n* **lista de pestañas del navegador**\n\n * Título: Pestañas de lista\n * Descripción: Lista de pestañas del navegador\n * Parámetros: Ninguno\n * Solo lectura: **verdadero**\n\n* **pestaña\\_del\\_navegador\\_nueva**\n\n * Título: Abrir una nueva pestaña\n * Descripción: Abre una nueva pestaña\n * Parámetros:\n * `url` (cadena, opcional): URL a la que se accederá en la nueva pestaña. Si no se proporciona, la nueva pestaña estará en blanco.\n * Solo lectura: **verdadero**\n\n* **selección de pestaña del navegador**\n\n * Título: Seleccionar una pestaña\n * Descripción: Seleccionar una pestaña por índice\n * Parámetros:\n * `index` (número): El índice de la pestaña a seleccionar\n * Solo lectura: **verdadero**\n\n* **cerrar pestaña del navegador**\n\n * Título: Cerrar una pestaña\n * Descripción: Cerrar una pestaña\n * Parámetros:\n * `index` (número, opcional): El índice de la pestaña que se cerrará. Cierra la pestaña actual si no se proporciona.\n * Solo lectura: **falso**\n\n* **prueba de dramaturgo generada por el navegador**\n\n * Título: Generar una prueba de dramaturgo\n * Descripción: Generar una prueba de dramaturgo para un escenario determinado\n * Parámetros:\n * `name` (cadena): el nombre de la prueba\n * `description` (cadena): La descripción de la prueba\n * `steps` (matriz): Los pasos de la prueba\n * Solo lectura: **verdadero**\n\n* **captura de pantalla del navegador**\n\n * Título: Tomar una captura de pantalla\n * Descripción: Toma una captura de pantalla de la página actual\n * Parámetros: Ninguno\n * Solo lectura: **verdadero**\n\n* **mover\\_ratón\\_pantalla\\_navegador**\n\n * Título: Mover el ratón\n * Descripción: Mueve el ratón a una posición determinada\n * Parámetros:\n * `element` (cadena): Descripción del elemento legible por humanos que se utiliza para obtener permiso para interactuar con el elemento\n * `x` (número): coordenada X\n * `y` (número): coordenada Y\n * Solo lectura: **verdadero**\n\n* **clic en la pantalla del navegador**\n\n * Título: Haga clic\n * Descripción: Haga clic con el botón izquierdo del ratón.\n * Parámetros:\n * `element` (cadena): Descripción del elemento legible por humanos que se utiliza para obtener permiso para interactuar con el elemento\n * `x` (número): coordenada X\n * `y` (número): coordenada Y\n * Solo lectura: **falso**\n\n* **arrastrar la pantalla del navegador**\n\n * Título: Arrastre el ratón\n * Descripción: Arrastre el botón izquierdo del ratón\n * Parámetros:\n * `element` (cadena): Descripción del elemento legible por humanos que se utiliza para obtener permiso para interactuar con el elemento\n * `startX` (número): Coordenada X de inicio\n * `startY` (número): Coordenada Y de inicio\n * `endX` (número): Coordenada final X\n * `endY` (número): Coordenada Y final\n * Solo lectura: **falso**\n\n* **tipo de pantalla del navegador**\n\n * Título: Escriba el texto\n * Descripción: Escriba texto\n * Parámetros:\n * `text` (cadena): Texto a escribir en el elemento\n * `submit` (booleano, opcional): si se debe enviar el texto ingresado (presione Enter después)\n * Solo lectura: **falso**\n\n* **tecla\\_pulsar\\_del\\_navegador**\n\n * Título: Presiona una tecla\n * Descripción: Presione una tecla del teclado\n * Parámetros:\n * `key` (cadena): Nombre de la tecla a presionar o un carácter a generar, como `ArrowLeft` o `a`\n * Solo lectura: **falso**\n\n* **espera del navegador**\n\n * Título: Espera\n * Descripción: Esperar a que aparezca o desaparezca el texto o que transcurra un tiempo específico\n * Parámetros:\n * `time` (número, opcional): el tiempo de espera en segundos\n * `text` (cadena, opcional): el texto que se debe esperar\n * `textGone` (cadena, opcional): el texto que se debe esperar a que desaparezca\n * Solo lectura: **verdadero**\n\n* **carga de archivos del navegador**\n\n * Título: Subir archivos\n * Descripción: Subir uno o varios archivos\n * Parámetros:\n * `paths` (matriz): Las rutas absolutas de los archivos que se van a cargar. Pueden ser uno o varios archivos.\n * Solo lectura: **falso**\n\n* **diálogo del identificador del navegador**\n\n * Título: Manejar un diálogo\n * Descripción: Manejar un diálogo\n * Parámetros:\n * `accept` (booleano): si se debe aceptar el diálogo.\n * `promptText` (cadena, opcional): el texto del mensaje en caso de un cuadro de diálogo de mensaje.\n * Solo lectura: **falso**","Spanish","es-ES",{"_36":62,"_40":66,"_45":67},{"_38":63,"_40":64,"_42":65},"ビジョンモデルやスクリーンショットを必要とせずに、構造化されたアクセシビリティスナップショットを通じて LLM が Web ページと対話できるようにするモデルコンテキストプロトコルサーバー。","Playwright MCP サーバー","## 劇作家MCP\n\n[Playwright](https://playwright.dev)を用いたブラウザ自動化機能を提供するモデルコンテキストプロトコル（MCP）サーバー。このサーバーにより、LLMは構造化されたアクセシビリティスナップショットを通じてWebページと対話できるようになり、スクリーンショットや視覚的に調整されたモデルを必要としません。\n\n### 主な特徴\n\n* **高速かつ軽量**。ピクセルベースの入力ではなく、Playwright のアクセシビリティツリーを使用します。\n* **LLM 対応**。ビジョンモデルは必要なく、純粋に構造化データ上で動作します。\n* **決定論的なツールアプリケーション**。スクリーンショットベースのアプローチでよくある曖昧さを回避します。\n\n### 要件\n\n* Node.js 18以降\n* VS Code、Cursor、Windsurf、Claude Desktop、その他のMCPクライアント\n\n### はじめる\n\nまず、クライアントにPlaywright MCPサーバーをインストールします。典型的な構成は以下のとおりです。\n\n```js\n{\n \"mcpServers\": {\n \"playwright\": {\n \"command\": \"npx\",\n \"args\": [\n \"@playwright/mcp@latest\"\n ]\n }\n }\n}\n```\n\nVS Code CLI を使用して Playwright MCP サーバーをインストールすることもできます。\n\n```bash\n# For VS Code\ncode --add-mcp '{\"name\":\"playwright\",\"command\":\"npx\",\"args\":[\"@playwright/mcp@latest\"]}'\n```\n\nインストール後、Playwright MCP サーバーは VS Code 内の GitHub Copilot エージェントで使用できるようになります。\n\n`Cursor Settings` -> `MCP` -> `Add new MCP Server`に進みます。お好みの名前を入力し、 `command`タイプを使用して`npx @playwright/mcp`コマンドを実行します。 `Edit`をクリックして設定を確認したり、コマンドのような引数を追加したりすることもできます。\n\n```js\n{\n \"mcpServers\": {\n \"playwright\": {\n \"command\": \"npx\",\n \"args\": [\n \"@playwright/mcp@latest\"\n ]\n }\n }\n}\n```\n\nWindsuff MCP[ドキュメント](https://docs.windsurf.com/windsurf/cascade/mcp)に従ってください。以下の設定を使用してください。\n\n```js\n{\n \"mcpServers\": {\n \"playwright\": {\n \"command\": \"npx\",\n \"args\": [\n \"@playwright/mcp@latest\"\n ]\n }\n }\n}\n```\n\nMCP インストール[ガイド](https://modelcontextprotocol.io/quickstart/user)に従って、次の構成を使用します。\n\n```js\n{\n \"mcpServers\": {\n \"playwright\": {\n \"command\": \"npx\",\n \"args\": [\n \"@playwright/mcp@latest\"\n ]\n }\n }\n}\n```\n\n### 構成\n\nPlaywright MCPサーバーは以下の引数をサポートしています。これらは上記のJSON設定の`\"args\"`リストの一部として指定できます。\n\n```\n> npx @playwright/mcp@latest --help\n --allowed-origins semicolon-separated list of origins to allow the\n browser to request. Default is to allow all.\n --blocked-origins semicolon-separated list of origins to block the\n browser from requesting. Blocklist is evaluated\n before allowlist. If used without the allowlist,\n requests not matching the blocklist are still\n allowed.\n --block-service-workers block service workers\n --browser browser or chrome channel to use, possible\n values: chrome, firefox, webkit, msedge.\n --caps comma-separated list of capabilities to enable,\n possible values: tabs, pdf, history, wait, files,\n install. Default is all.\n --cdp-endpoint CDP endpoint to connect to.\n --config path to the configuration file.\n --device device to emulate, for example: \"iPhone 15\"\n --executable-path path to the browser executable.\n --headless run browser in headless mode, headed by default\n --host host to bind server to. Default is localhost. Use\n 0.0.0.0 to bind to all interfaces.\n --ignore-https-errors ignore https errors\n --isolated keep the browser profile in memory, do not save\n it to disk.\n --no-image-responses do not send image responses to the client.\n --no-sandbox disable the sandbox for all process types that\n are normally sandboxed.\n --output-dir path to the directory for output files.\n --port port to listen on for SSE transport.\n --proxy-bypass comma-separated domains to bypass proxy, for\n example \".com,chromium.org,.domain.com\"\n --proxy-server specify proxy server, for example\n \"http://myproxy:3128\" or \"socks5://myproxy:8080\"\n --save-trace Whether to save the Playwright Trace of the\n session into the output directory.\n --storage-state path to the storage state file for isolated\n sessions.\n --user-agent specify user agent string\n --user-data-dir path to the user data directory. If not\n specified, a temporary directory will be created.\n --viewport-size specify browser viewport size in pixels, for\n example \"1280, 720\"\n --vision Run server that uses screenshots (Aria snapshots\n are used by default)\n```\n\n### ユーザープロフィール\n\nPlaywright MCP は、通常のブラウザ (デフォルト) のように永続プロファイルを使用して実行することも、テストセッション用の分離されたコンテキストで実行することもできます。\n\n**永続的なプロファイル**\n\nログイン情報はすべて永続プロファイルに保存されます。オフライン状態をクリアしたい場合は、セッション間でこのプロファイルを削除できます。永続プロファイルは以下の場所に保存されており、 `--user-data-dir`引数で上書きできます。\n\n```bash\n# Windows\n%USERPROFILE%\\AppData\\Local\\ms-playwright\\mcp-{channel}-profile\n\n# macOS\n- ~/Library/Caches/ms-playwright/mcp-{channel}-profile\n\n# Linux\n- ~/.cache/ms-playwright/mcp-{channel}-profile\n```\n\n**孤立した**\n\n分離モードでは、各セッションは分離プロファイルで開始されます。MCPにブラウザを閉じるよう指示するたびにセッションが閉じられ、そのセッションのすべてのストレージ状態が失われます。設定の`contextOptions`または`--storage-state`引数を使用して、ブラウザに初期のストレージ状態を提供できます。ストレージ状態の詳細については、[こちらを](https://playwright.dev/docs/auth)ご覧ください。\n\n```js\n{\n \"mcpServers\": {\n \"playwright\": {\n \"command\": \"npx\",\n \"args\": [\n \"@playwright/mcp@latest\",\n \"--isolated\",\n \"--storage-state={path/to/storage.json}\n ]\n }\n }\n}\n```\n\n### 設定ファイル\n\nPlaywright MCPサーバーはJSON設定ファイルを使用して設定できます。設定ファイルは、 `--config`コマンドラインオプションを使用して指定できます。\n\n```bash\nnpx @playwright/mcp@latest --config path/to/config.json\n```\n\n```typescript\n{\n // Browser configuration\n browser?: {\n // Browser type to use (chromium, firefox, or webkit)\n browserName?: 'chromium' | 'firefox' | 'webkit';\n\n // Keep the browser profile in memory, do not save it to disk.\n isolated?: boolean;\n\n // Path to user data directory for browser profile persistence\n userDataDir?: string;\n\n // Browser launch options (see Playwright docs)\n // @see https://playwright.dev/docs/api/class-browsertype#browser-type-launch\n launchOptions?: {\n channel?: string; // Browser channel (e.g. 'chrome')\n headless?: boolean; // Run in headless mode\n executablePath?: string; // Path to browser executable\n // ... other Playwright launch options\n };\n\n // Browser context options\n // @see https://playwright.dev/docs/api/class-browser#browser-new-context\n contextOptions?: {\n viewport?: { width: number, height: number };\n // ... other Playwright context options\n };\n\n // CDP endpoint for connecting to existing browser\n cdpEndpoint?: string;\n\n // Remote Playwright server endpoint\n remoteEndpoint?: string;\n },\n\n // Server configuration\n server?: {\n port?: number; // Port to listen on\n host?: string; // Host to bind to (default: localhost)\n },\n\n // List of enabled capabilities\n capabilities?: Array<\n 'core' | // Core browser automation\n 'tabs' | // Tab management\n 'pdf' | // PDF generation\n 'history' | // Browser history\n 'wait' | // Wait utilities\n 'files' | // File handling\n 'install' | // Browser installation\n 'testing' // Testing\n >;\n\n // Enable vision mode (screenshots instead of accessibility snapshots)\n vision?: boolean;\n\n // Directory for output files\n outputDir?: string;\n\n // Network configuration\n network?: {\n // List of origins to allow the browser to request. Default is to allow all. Origins matching both `allowedOrigins` and `blockedOrigins` will be blocked.\n allowedOrigins?: string[];\n\n // List of origins to block the browser to request. Origins matching both `allowedOrigins` and `blockedOrigins` will be blocked.\n blockedOrigins?: string[];\n };\n \n /**\n * Do not send image responses to the client.\n */\n noImageResponses?: boolean;\n}\n```\n\n### スタンドアロンMCPサーバー\n\nディスプレイのないシステム上または IDE のワーカープロセスからヘッダー付きブラウザーを実行する場合は、DISPLAY を指定した環境から MCP サーバーを実行し、 `--port`フラグを渡して SSE トランスポートを有効にします。\n\n```bash\nnpx @playwright/mcp@latest --port 8931\n```\n\n次に、MCP クライアント構成で、SSE エンドポイントへの`url`を設定します。\n\n```js\n{\n \"mcpServers\": {\n \"playwright\": {\n \"url\": \"http://localhost:8931/sse\"\n }\n }\n}\n```\n\n\\*\\*注意:\\*\\*現時点では、Docker 実装ではヘッドレス Chromium のみがサポートされています。\n\n```js\n{\n \"mcpServers\": {\n \"playwright\": {\n \"command\": \"docker\",\n \"args\": [\"run\", \"-i\", \"--rm\", \"--init\", \"--pull=always\", \"mcr.microsoft.com/playwright/mcp\"]\n }\n }\n}\n```\n\nDocker イメージを自分でビルドできます。\n\n```\ndocker build -t mcr.microsoft.com/playwright/mcp .\n```\n\n```js\nimport http from 'http';\n\nimport { createConnection } from '@playwright/mcp';\nimport { SSEServerTransport } from '@modelcontextprotocol/sdk/server/sse.js';\n\nhttp.createServer(async (req, res) => {\n // ...\n\n // Creates a headless Playwright MCP server with SSE transport\n const connection = await createConnection({ browser: { launchOptions: { headless: true } } });\n const transport = new SSEServerTransport('/messages', res);\n await connection.connect(transport);\n\n // ...\n});\n```\n\n### ツール\n\nツールは次の 2 つのモードで使用できます。\n\n1. **スナップショットモード**（デフォルト）：アクセシビリティスナップショットを使用してパフォーマンスと信頼性を向上します\n2. **ビジョンモード**: スクリーンショットを使用して視覚的なインタラクションを行います\n\nビジョンモードを使用するには、サーバーの起動時に`--vision`フラグを追加します。\n\n```js\n{\n \"mcpServers\": {\n \"playwright\": {\n \"command\": \"npx\",\n \"args\": [\n \"@playwright/mcp@latest\",\n \"--vision\"\n ]\n }\n }\n}\n```\n\nビジョンモードは、提供されたスクリーンショットに基づいて、XY 座標空間を使用して要素と対話できるコンピューター使用モデルで最も効果的に機能します。\n\n* **ブラウザのスナップショット**\n\n * タイトル: ページスナップショット\n * 説明: 現在のページのアクセシビリティスナップショットをキャプチャします。これはスクリーンショットよりも優れています。\n * パラメータ: なし\n * 読み取り専用: **true**\n\n* **ブラウザクリック**\n\n * タイトル: クリック\n * 説明: Webページでクリックを実行する\n * パラメータ:\n * `element` (文字列): 要素と対話するための許可を得るために使用される、人間が読める要素の説明\n * `ref` (文字列): ページスナップショットからの正確なターゲット要素参照\n * 読み取り専用: **false**\n\n* **ブラウザドラッグ**\n\n * タイトル: マウスをドラッグ\n * 説明: 2つの要素間でドラッグアンドドロップを実行します\n * パラメータ:\n * `startElement` (文字列): 要素と対話するための許可を取得するために使用される、人間が読めるソース要素の説明\n * `startRef` (文字列): ページスナップショットからの正確なソース要素参照\n * `endElement` (文字列): 要素と対話するための許可を取得するために使用される、人間が読めるターゲット要素の説明\n * `endRef` (文字列): ページスナップショットからの正確なターゲット要素参照\n * 読み取り専用: **false**\n\n* **ブラウザホバー**\n\n * タイトル: ホバーマウス\n * 説明: ページ上の要素にマウスを移動します\n * パラメータ:\n * `element` (文字列): 要素と対話するための許可を得るために使用される、人間が読める要素の説明\n * `ref` (文字列): ページスナップショットからの正確なターゲット要素参照\n * 読み取り専用: **true**\n\n* **ブラウザの種類**\n\n * タイトル: テキストを入力\n * 説明: 編集可能な要素にテキストを入力します\n * パラメータ:\n * `element` (文字列): 要素と対話するための許可を得るために使用される、人間が読める要素の説明\n * `ref` (文字列): ページスナップショットからの正確なターゲット要素参照\n * `text` (文字列): 要素に入力するテキスト\n * `submit` （ブール値、オプション）：入力したテキストを送信するかどうか（送信後にEnterキーを押す）\n * `slowly` (boolean, オプション): 一度に1文字ずつ入力するかどうか。ページ内のキーハンドラをトリガーするのに便利です。デフォルトでは、テキスト全体が一度に入力されます。\n * 読み取り専用: **false**\n\n* **ブラウザ選択オプション**\n\n * タイトル: オプションを選択\n * 説明: ドロップダウンからオプションを選択します\n * パラメータ:\n * `element` (文字列): 要素と対話するための許可を得るために使用される、人間が読める要素の説明\n * `ref` (文字列): ページスナップショットからの正確なターゲット要素参照\n * `values` (配列): ドロップダウンで選択する値の配列。単一の値または複数の値を指定できます。\n * 読み取り専用: **false**\n\n* **ブラウザのキーを押す**\n\n * タイトル: キーを押す\n * 説明: キーボードのキーを押します\n * パラメータ:\n * `key` (文字列): 押すキーの名前、または生成する`a` （例： `ArrowLeft` 、\n * 読み取り専用: **false**\n\n* **ブラウザの待機時間**\n\n * タイトル: 待つ\n * 説明: テキストが表示されたり消えたりするか、指定された時間が経過するのを待ちます\n * パラメータ:\n * `time` （数値、オプション）：待機時間（秒）\n * `text` （文字列、オプション）：待機するテキスト\n * `textGone` (文字列、オプション): 消えるのを待つテキスト\n * 読み取り専用: **true**\n\n* **ブラウザファイルアップロード**\n\n * タイトル: ファイルをアップロード\n * 説明: 1つまたは複数のファイルをアップロードします\n * パラメータ:\n * `paths` (配列): アップロードするファイルへの絶対パス。1つのファイルでも複数のファイルでも構いません。\n * 読み取り専用: **false**\n\n* **ブラウザハンドルダイアログ**\n\n * タイトル: ダイアログを処理する\n * 説明: ダイアログを処理する\n * パラメータ:\n * `accept` (boolean): ダイアログを受け入れるかどうか。\n * `promptText` (文字列、オプション): プロンプトダイアログの場合のプロンプトのテキスト。\n * 読み取り専用: **false**\n\n* **ブラウザナビゲート**\n\n * タイトル: URL に移動する\n * 説明: URL に移動する\n * パラメータ:\n * `url` (文字列): 移動するURL\n * 読み取り専用: **false**\n\n* **ブラウザの戻る**\n\n * タイトル: 戻る\n * 説明: 前のページに戻る\n * パラメータ: なし\n * 読み取り専用: **true**\n\n* **ブラウザで進む**\n\n * タイトル: 前進\n * 説明: 次のページへ進む\n * パラメータ: なし\n * 読み取り専用: **true**\n\n* **ブラウザのスクリーンショット**\n\n * タイトル: スクリーンショットを撮る\n * 説明: 現在のページのスクリーンショットを撮ります。スクリーンショットに基づいてアクションを実行することはできません。アクションには browser\\_snapshot を使用してください。\n * パラメータ:\n * `raw` (ブール値、オプション): 圧縮せずにPNG形式で返すかどうか。デフォルトはfalseで、JPEG画像を返します。\n * `filename` (文字列, オプション): スクリーンショットを保存するファイル名。指定されていない場合は、デフォルトで`page-{timestamp}.{png|jpeg}`になります。\n * `element` (文字列, オプション): 要素のスクリーンショットの許可を得るために使用される、人間が読める形式の要素説明。指定されていない場合は、ビューポートのスクリーンショットが撮影されます。element を指定する場合は、ref も指定する必要があります。\n * `ref` (文字列, オプション): ページスナップショットから取得する正確なターゲット要素の参照。指定されていない場合は、ビューポートのスクリーンショットが取得されます。ref を指定する場合は、要素も指定する必要があります。\n * 読み取り専用: **true**\n\n* **ブラウザPDF保存**\n\n * タイトル: PDFとして保存\n * 説明: ページをPDFとして保存\n * パラメータ:\n * `filename` (文字列、オプション): PDFを保存するファイル名。指定されていない場合は、デフォルトで`page-{timestamp}.pdf`になります。\n * 読み取り専用: **true**\n\n* **ブラウザネットワークリクエスト**\n\n * タイトル: ネットワークリクエストの一覧\n * 説明: ページの読み込み以降のすべてのネットワークリクエストを返します\n * パラメータ: なし\n * 読み取り専用: **true**\n\n* **ブラウザコンソールメッセージ**\n\n * タイトル: コンソールメッセージを取得する\n * 説明: すべてのコンソールメッセージを返します\n * パラメータ: なし\n * 読み取り専用: **true**\n\n* **ブラウザのインストール**\n\n * タイトル: 設定で指定されたブラウザをインストールする\n * 説明: 設定で指定されたブラウザをインストールします。ブラウザがインストールされていないというエラーが発生した場合にこれを呼び出します。\n * パラメータ: なし\n * 読み取り専用: **false**\n\n* **ブラウザを閉じる**\n\n * タイトル: ブラウザを閉じる\n * 説明: ページを閉じる\n * パラメータ: なし\n * 読み取り専用: **true**\n\n* **ブラウザのサイズ変更**\n\n * タイトル: ブラウザウィンドウのサイズを変更する\n * 説明: ブラウザウィンドウのサイズを変更する\n * パラメータ:\n * `width` （数値）：ブラウザウィンドウの幅\n * `height` （数値）：ブラウザウィンドウの高さ\n * 読み取り専用: **true**\n\n* **ブラウザタブリスト**\n\n * タイトル: リストタブ\n * 説明: ブラウザのタブを一覧表示する\n * パラメータ: なし\n * 読み取り専用: **true**\n\n* **ブラウザタブ新規**\n\n * タイトル: 新しいタブを開く\n * 説明: 新しいタブを開く\n * パラメータ:\n * `url` (文字列、オプション): 新しいタブで移動するURL。指定しない場合、新しいタブは空白になります。\n * 読み取り専用: **true**\n\n* **ブラウザタブ選択**\n\n * タイトル: タブを選択\n * 説明: インデックスでタブを選択する\n * パラメータ:\n * `index` （数値）：選択するタブのインデックス\n * 読み取り専用: **true**\n\n* **ブラウザタブを閉じる**\n\n * タイトル: タブを閉じる\n * 説明: タブを閉じる\n * パラメータ:\n * `index` (数値, オプション): 閉じるタブのインデックス。指定されていない場合は現在のタブを閉じます。\n * 読み取り専用: **false**\n\n* **ブラウザ生成プレイライトテスト**\n\n * タイトル: Playwright テストを生成する\n * 説明: 指定されたシナリオの Playwright テストを生成します\n * パラメータ:\n * `name` (文字列): テストの名前\n * `description` (文字列): テストの説明\n * `steps` （配列）：テストのステップ\n * 読み取り専用: **true**\n\n* **ブラウザのスクリーンキャプチャ**\n\n * タイトル: スクリーンショットを撮る\n * 説明: 現在のページのスクリーンショットを撮ります\n * パラメータ: なし\n * 読み取り専用: **true**\n\n* **ブラウザ画面上のマウス移動**\n\n * タイトル: マウスを動かす\n * 説明: マウスを指定された位置に移動する\n * パラメータ:\n * `element` (文字列): 要素と対話するための許可を得るために使用される、人間が読める要素の説明\n * `x` (数値): X座標\n * `y` (数値): Y座標\n * 読み取り専用: **true**\n\n* **ブラウザ画面のクリック**\n\n * タイトル: クリック\n * 説明: マウスの左ボタンをクリック\n * パラメータ:\n * `element` (文字列): 要素と対話するための許可を得るために使用される、人間が読める要素の説明\n * `x` (数値): X座標\n * `y` (数値): Y座標\n * 読み取り専用: **false**\n\n* **ブラウザ画面ドラッグ**\n\n * タイトル: マウスをドラッグ\n * 説明: 左マウスボタンをドラッグ\n * パラメータ:\n * `element` (文字列): 要素と対話するための許可を得るために使用される、人間が読める要素の説明\n * `startX` (数値): 開始X座標\n * `startY` (数値): 開始Y座標\n * `endX` (数値): 終了X座標\n * `endY` (数値): 終了Y座標\n * 読み取り専用: **false**\n\n* **ブラウザ画面タイプ**\n\n * タイトル: テキストを入力\n * 説明: テキストを入力\n * パラメータ:\n * `text` (文字列): 要素に入力するテキスト\n * `submit` （ブール値、オプション）：入力したテキストを送信するかどうか（送信後にEnterキーを押す）\n * 読み取り専用: **false**\n\n* **ブラウザのキーを押す**\n\n * タイトル: キーを押す\n * 説明: キーボードのキーを押します\n * パラメータ:\n * `key` (文字列): 押すキーの名前、または生成する`a` （例： `ArrowLeft` 、\n * 読み取り専用: **false**\n\n* **ブラウザの待機時間**\n\n * タイトル: 待つ\n * 説明: テキストが表示されたり消えたりするか、指定された時間が経過するのを待ちます\n * パラメータ:\n * `time` （数値、オプション）：待機時間（秒）\n * `text` （文字列、オプション）：待機するテキスト\n * `textGone` (文字列、オプション): 消えるのを待つテキスト\n * 読み取り専用: **true**\n\n* **ブラウザファイルアップロード**\n\n * タイトル: ファイルをアップロード\n * 説明: 1つまたは複数のファイルをアップロードします\n * パラメータ:\n * `paths` (配列): アップロードするファイルへの絶対パス。1つのファイルでも複数のファイルでも構いません。\n * 読み取り専用: **false**\n\n* **ブラウザハンドルダイアログ**\n\n * タイトル: ダイアログを処理する\n * 説明: ダイアログを処理する\n * パラメータ:\n * `accept` (boolean): ダイアログを受け入れるかどうか。\n * `promptText` (文字列、オプション): プロンプトダイアログの場合のプロンプトのテキスト。\n * 読み取り専用: **false**","Japanese","ja-JP",{"_36":69,"_40":73,"_45":74},{"_38":70,"_40":71,"_42":72},"비전 모델이나 스크린샷이 필요 없이 구조화된 접근성 스냅샷을 통해 LLM이 웹 페이지와 상호 작용할 수 있도록 하는 모델 컨텍스트 프로토콜 서버입니다.","극작가 MCP 서버","## 극작가 MCP\n\n[Playwright를](https://playwright.dev) 사용하여 브라우저 자동화 기능을 제공하는 모델 컨텍스트 프로토콜(MCP) 서버입니다. 이 서버를 사용하면 LLM이 구조화된 접근성 스냅샷을 통해 웹 페이지와 상호 작용할 수 있으므로 스크린샷이나 시각적으로 조정된 모델이 필요하지 않습니다.\n\n### 주요 특징\n\n* **빠르고 가볍습니다** . 픽셀 기반 입력 방식이 아닌 Playwright의 접근성 트리를 사용합니다.\n* **LLM 친화적입니다** . 비전 모델이 필요 없으며, 구조화된 데이터만을 기반으로 운영됩니다.\n* **결정론적 도구 적용** . 스크린샷 기반 접근 방식에서 흔히 발생하는 모호성을 방지합니다.\n\n### 요구 사항\n\n* Node.js 18 이상\n* VS Code, Cursor, Windsurf, Claude Desktop 또는 기타 MCP 클라이언트\n\n### 시작하기\n\n먼저 클라이언트와 함께 Playwright MCP 서버를 설치하세요. 일반적인 구성은 다음과 같습니다.\n\n지엑스피1\n\nVS Code CLI를 사용하여 Playwright MCP 서버를 설치할 수도 있습니다.\n\n```bash\n# For VS Code\ncode --add-mcp '{\"name\":\"playwright\",\"command\":\"npx\",\"args\":[\"@playwright/mcp@latest\"]}'\n```\n\n설치 후 Playwright MCP 서버를 VS Code의 GitHub Copilot 에이전트와 함께 사용할 수 있습니다.\n\n`Cursor Settings` -> `MCP` -> `Add new MCP Server` 로 이동합니다. 원하는 이름을 입력하고, `npx @playwright/mcp` 명령어와 함께 `command` 유형을 사용합니다. ' `Edit` 클릭하여 구성을 확인하거나 명령어와 같은 인수를 추가할 수도 있습니다.\n\n```js\n{\n \"mcpServers\": {\n \"playwright\": {\n \"command\": \"npx\",\n \"args\": [\n \"@playwright/mcp@latest\"\n ]\n }\n }\n}\n```\n\nWindsuff MCP [설명서를](https://docs.windsurf.com/windsurf/cascade/mcp) 따르세요. 다음 구성을 사용하세요.\n\n```js\n{\n \"mcpServers\": {\n \"playwright\": {\n \"command\": \"npx\",\n \"args\": [\n \"@playwright/mcp@latest\"\n ]\n }\n }\n}\n```\n\nMCP 설치 [가이드를](https://modelcontextprotocol.io/quickstart/user) 따르고 다음 구성을 사용하세요.\n\n```js\n{\n \"mcpServers\": {\n \"playwright\": {\n \"command\": \"npx\",\n \"args\": [\n \"@playwright/mcp@latest\"\n ]\n }\n }\n}\n```\n\n### 구성\n\nPlaywright MCP 서버는 다음 인수를 지원합니다. 위 JSON 구성에서 `\"args\"` 목록의 일부로 제공할 수 있습니다.\n\n```\n> npx @playwright/mcp@latest --help\n --allowed-origins semicolon-separated list of origins to allow the\n browser to request. Default is to allow all.\n --blocked-origins semicolon-separated list of origins to block the\n browser from requesting. Blocklist is evaluated\n before allowlist. If used without the allowlist,\n requests not matching the blocklist are still\n allowed.\n --block-service-workers block service workers\n --browser browser or chrome channel to use, possible\n values: chrome, firefox, webkit, msedge.\n --caps comma-separated list of capabilities to enable,\n possible values: tabs, pdf, history, wait, files,\n install. Default is all.\n --cdp-endpoint CDP endpoint to connect to.\n --config path to the configuration file.\n --device device to emulate, for example: \"iPhone 15\"\n --executable-path path to the browser executable.\n --headless run browser in headless mode, headed by default\n --host host to bind server to. Default is localhost. Use\n 0.0.0.0 to bind to all interfaces.\n --ignore-https-errors ignore https errors\n --isolated keep the browser profile in memory, do not save\n it to disk.\n --no-image-responses do not send image responses to the client.\n --no-sandbox disable the sandbox for all process types that\n are normally sandboxed.\n --output-dir path to the directory for output files.\n --port port to listen on for SSE transport.\n --proxy-bypass comma-separated domains to bypass proxy, for\n example \".com,chromium.org,.domain.com\"\n --proxy-server specify proxy server, for example\n \"http://myproxy:3128\" or \"socks5://myproxy:8080\"\n --save-trace Whether to save the Playwright Trace of the\n session into the output directory.\n --storage-state path to the storage state file for isolated\n sessions.\n --user-agent specify user agent string\n --user-data-dir path to the user data directory. If not\n specified, a temporary directory will be created.\n --viewport-size specify browser viewport size in pixels, for\n example \"1280, 720\"\n --vision Run server that uses screenshots (Aria snapshots\n are used by default)\n```\n\n### 사용자 프로필\n\n일반 브라우저(기본값)처럼 지속적인 프로필로 Playwright MCP를 실행하거나 테스트 세션을 위한 격리된 컨텍스트에서 실행할 수 있습니다.\n\n**영구 프로필**\n\n모든 로그인 정보는 영구 프로필에 저장되며, 오프라인 상태를 해제하려면 세션 간에 삭제할 수 있습니다. 영구 프로필은 다음 위치에 있으며, `--user-data-dir` 인수를 사용하여 재정의할 수 있습니다.\n\n```bash\n# Windows\n%USERPROFILE%\\AppData\\Local\\ms-playwright\\mcp-{channel}-profile\n\n# macOS\n- ~/Library/Caches/ms-playwright/mcp-{channel}-profile\n\n# Linux\n- ~/.cache/ms-playwright/mcp-{channel}-profile\n```\n\n**외딴**\n\n격리 모드에서는 각 세션이 격리된 프로필에서 시작됩니다. MCP에 브라우저를 닫도록 요청할 때마다 세션이 닫히고 해당 세션의 모든 저장소 상태가 손실됩니다. 구성의 `contextOptions` 또는 `--storage-state` 인수를 통해 브라우저에 초기 저장소 상태를 제공할 수 있습니다. 저장소 상태에 대한 자세한 내용은 [여기를](https://playwright.dev/docs/auth) 참조하세요.\n\n```js\n{\n \"mcpServers\": {\n \"playwright\": {\n \"command\": \"npx\",\n \"args\": [\n \"@playwright/mcp@latest\",\n \"--isolated\",\n \"--storage-state={path/to/storage.json}\n ]\n }\n }\n}\n```\n\n### 구성 파일\n\nPlaywright MCP 서버는 JSON 구성 파일을 사용하여 구성할 수 있습니다. `--config` 명령줄 옵션을 사용하여 구성 파일을 지정할 수 있습니다.\n\n```bash\nnpx @playwright/mcp@latest --config path/to/config.json\n```\n\n```typescript\n{\n // Browser configuration\n browser?: {\n // Browser type to use (chromium, firefox, or webkit)\n browserName?: 'chromium' | 'firefox' | 'webkit';\n\n // Keep the browser profile in memory, do not save it to disk.\n isolated?: boolean;\n\n // Path to user data directory for browser profile persistence\n userDataDir?: string;\n\n // Browser launch options (see Playwright docs)\n // @see https://playwright.dev/docs/api/class-browsertype#browser-type-launch\n launchOptions?: {\n channel?: string; // Browser channel (e.g. 'chrome')\n headless?: boolean; // Run in headless mode\n executablePath?: string; // Path to browser executable\n // ... other Playwright launch options\n };\n\n // Browser context options\n // @see https://playwright.dev/docs/api/class-browser#browser-new-context\n contextOptions?: {\n viewport?: { width: number, height: number };\n // ... other Playwright context options\n };\n\n // CDP endpoint for connecting to existing browser\n cdpEndpoint?: string;\n\n // Remote Playwright server endpoint\n remoteEndpoint?: string;\n },\n\n // Server configuration\n server?: {\n port?: number; // Port to listen on\n host?: string; // Host to bind to (default: localhost)\n },\n\n // List of enabled capabilities\n capabilities?: Array<\n 'core' | // Core browser automation\n 'tabs' | // Tab management\n 'pdf' | // PDF generation\n 'history' | // Browser history\n 'wait' | // Wait utilities\n 'files' | // File handling\n 'install' | // Browser installation\n 'testing' // Testing\n >;\n\n // Enable vision mode (screenshots instead of accessibility snapshots)\n vision?: boolean;\n\n // Directory for output files\n outputDir?: string;\n\n // Network configuration\n network?: {\n // List of origins to allow the browser to request. Default is to allow all. Origins matching both `allowedOrigins` and `blockedOrigins` will be blocked.\n allowedOrigins?: string[];\n\n // List of origins to block the browser to request. Origins matching both `allowedOrigins` and `blockedOrigins` will be blocked.\n blockedOrigins?: string[];\n };\n \n /**\n * Do not send image responses to the client.\n */\n noImageResponses?: boolean;\n}\n```\n\n### 독립형 MCP 서버\n\n디스플레이가 없는 시스템이나 IDE의 작업자 프로세스에서 헤더 브라우저를 실행하는 경우 DISPLAY가 있는 환경에서 MCP 서버를 실행하고 `--port` 플래그를 전달하여 SSE 전송을 활성화합니다.\n\n```bash\nnpx @playwright/mcp@latest --port 8931\n```\n\n그런 다음 MCP 클라이언트 구성에서 `url` SSE 엔드포인트로 설정합니다.\n\n```js\n{\n \"mcpServers\": {\n \"playwright\": {\n \"url\": \"http://localhost:8931/sse\"\n }\n }\n}\n```\n\n**참고:** Docker 구현은 현재 헤드리스 크로미엄만 지원합니다.\n\n```js\n{\n \"mcpServers\": {\n \"playwright\": {\n \"command\": \"docker\",\n \"args\": [\"run\", \"-i\", \"--rm\", \"--init\", \"--pull=always\", \"mcr.microsoft.com/playwright/mcp\"]\n }\n }\n}\n```\n\nDocker 이미지를 직접 빌드할 수 있습니다.\n\n```\ndocker build -t mcr.microsoft.com/playwright/mcp .\n```\n\n```js\nimport http from 'http';\n\nimport { createConnection } from '@playwright/mcp';\nimport { SSEServerTransport } from '@modelcontextprotocol/sdk/server/sse.js';\n\nhttp.createServer(async (req, res) => {\n // ...\n\n // Creates a headless Playwright MCP server with SSE transport\n const connection = await createConnection({ browser: { launchOptions: { headless: true } } });\n const transport = new SSEServerTransport('/messages', res);\n await connection.connect(transport);\n\n // ...\n});\n```\n\n### 도구\n\n도구는 두 가지 모드로 제공됩니다.\n\n1. **스냅샷 모드** (기본값): 더 나은 성능과 안정성을 위해 접근성 스냅샷을 사용합니다.\n2. **비전 모드** : 시각적 기반 상호작용을 위해 스크린샷을 사용합니다.\n\nVision Mode를 사용하려면 서버를 시작할 때 `--vision` 플래그를 추가하세요.\n\n```js\n{\n \"mcpServers\": {\n \"playwright\": {\n \"command\": \"npx\",\n \"args\": [\n \"@playwright/mcp@latest\",\n \"--vision\"\n ]\n }\n }\n}\n```\n\n제공된 스크린샷을 기준으로, 비전 모드는 XY 좌표 공간을 사용하여 요소와 상호 작용할 수 있는 컴퓨터 사용 모델에서 가장 잘 작동합니다.\n\n* **브라우저 스냅샷**\n\n * 제목: 페이지 스냅샷\n * 설명: 현재 페이지의 접근성 스냅샷을 캡처합니다. 스크린샷보다 낫습니다.\n * 매개변수: 없음\n * 읽기 전용: **true**\n\n* **브라우저 클릭**\n\n * 제목: 클릭\n * 설명: 웹 페이지를 클릭합니다.\n * 매개변수:\n * `element` (문자열): 요소와 상호 작용할 수 있는 권한을 얻는 데 사용되는 사람이 읽을 수 있는 요소 설명\n * `ref` (문자열): 페이지 스냅샷의 정확한 대상 요소 참조\n * 읽기 전용: **false**\n\n* **브라우저 드래그**\n\n * 제목: 드래그 마우스\n * 설명: 두 요소 사이에서 드래그 앤 드롭을 수행합니다.\n * 매개변수:\n * `startElement` (문자열): 요소와 상호 작용할 수 있는 권한을 얻는 데 사용되는 사람이 읽을 수 있는 소스 요소 설명\n * `startRef` (문자열): 페이지 스냅샷의 정확한 소스 요소 참조\n * `endElement` (문자열): 요소와 상호 작용할 수 있는 권한을 얻는 데 사용되는 사람이 읽을 수 있는 대상 요소 설명\n * `endRef` (문자열): 페이지 스냅샷의 정확한 대상 요소 참조\n * 읽기 전용: **false**\n\n* **브라우저\\_호버**\n\n * 제목: 마우스를 올려주세요\n * 설명: 페이지의 요소 위에 마우스를 올려 놓으세요\n * 매개변수:\n * `element` (문자열): 요소와 상호 작용할 수 있는 권한을 얻는 데 사용되는 사람이 읽을 수 있는 요소 설명\n * `ref` (문자열): 페이지 스냅샷의 정확한 대상 요소 참조\n * 읽기 전용: **true**\n\n* **브라우저\\_유형**\n\n * 제목: 텍스트 입력\n * 설명: 편집 가능한 요소에 텍스트를 입력합니다.\n * 매개변수:\n * `element` (문자열): 요소와 상호 작용할 수 있는 권한을 얻는 데 사용되는 사람이 읽을 수 있는 요소 설명\n * `ref` (문자열): 페이지 스냅샷의 정확한 대상 요소 참조\n * `text` (문자열): 요소에 입력할 텍스트\n * `submit` (부울, 선택 사항): 입력한 텍스트를 제출할지 여부(제출 후 Enter 키 누름)\n * `slowly` (부울, 선택 사항): 한 번에 한 글자씩 입력할지 여부입니다. 페이지에서 키 핸들러를 트리거하는 데 유용합니다. 기본적으로 전체 텍스트가 한 번에 채워집니다.\n * 읽기 전용: **false**\n\n* **브라우저 선택 옵션**\n\n * 제목: 옵션 선택\n * 설명: 드롭다운에서 옵션을 선택하세요\n * 매개변수:\n * `element` (문자열): 요소와 상호 작용할 수 있는 권한을 얻는 데 사용되는 사람이 읽을 수 있는 요소 설명\n * `ref` (문자열): 페이지 스냅샷의 정확한 대상 요소 참조\n * `values` (배열): 드롭다운에서 선택할 값의 배열입니다. 단일 값 또는 여러 값일 수 있습니다.\n * 읽기 전용: **false**\n\n* **브라우저\\_누르기\\_키**\n\n * 제목: 키를 누르세요\n * 설명: 키보드의 키를 누르세요\n * 매개변수:\n * `key` (문자열): 누르려는 키의 이름 또는 생성할 문자(예: `ArrowLeft``a`\n * 읽기 전용: **false**\n\n* **브라우저 대기**\n\n * 제목 : 기다리다\n * 설명: 텍스트가 나타나거나 사라지거나 지정된 시간이 지날 때까지 기다립니다.\n * 매개변수:\n * `time` (숫자, 선택 사항): 대기 시간(초)\n * `text` (문자열, 선택 사항): 기다릴 텍스트\n * `textGone` (문자열, 선택 사항): 사라질 때까지 기다릴 텍스트\n * 읽기 전용: **true**\n\n* **브라우저 파일 업로드**\n\n * 제목: 파일 업로드\n * 설명: 하나 또는 여러 개의 파일을 업로드합니다.\n * 매개변수:\n * `paths` (배열): 업로드할 파일의 절대 경로입니다. 단일 파일 또는 여러 파일일 수 있습니다.\n * 읽기 전용: **false**\n\n* **브라우저\\_핸들\\_대화 상자**\n\n * 제목: 대화 처리\n * 설명: 대화 처리\n * 매개변수:\n * `accept` (boolean): 대화를 수락할지 여부.\n * `promptText` (문자열, 선택 사항): 프롬프트 대화 상자의 경우 프롬프트의 텍스트입니다.\n * 읽기 전용: **false**\n\n* **브라우저 탐색**\n\n * 제목: URL로 이동\n * 설명: URL로 이동합니다\n * 매개변수:\n * `url` (문자열): 이동할 URL\n * 읽기 전용: **false**\n\n* **브라우저\\_뒤로\\_탐색**\n\n * 제목: 돌아가기\n * 설명: 이전 페이지로 돌아가기\n * 매개변수: 없음\n * 읽기 전용: **true**\n\n* **브라우저\\_탐색\\_포워드**\n\n * 제목: 앞으로 나아가세요\n * 설명: 다음 페이지로 이동합니다.\n * 매개변수: 없음\n * 읽기 전용: **true**\n\n* **브라우저\\_스크린샷\\_찍기**\n\n * 제목: 스크린샷 찍기\n * 설명: 현재 페이지의 스크린샷을 찍습니다. 스크린샷을 기반으로 다른 작업을 수행할 수 없으므로, 작업에는 browser\\_snapshot을 사용하세요.\n * 매개변수:\n * `raw` (부울, 선택 사항): 압축하지 않고 PNG 형식으로 반환할지 여부입니다. 기본값은 false이며, JPEG 이미지를 반환합니다.\n * `filename` (문자열, 선택 사항): 스크린샷을 저장할 파일 이름입니다. 지정하지 않으면 기본값은 `page-{timestamp}.{png|jpeg}` .\n * `element` (문자열, 선택 사항): 사람이 읽을 수 있는 요소 설명으로, 요소의 스크린샷 권한을 얻는 데 사용됩니다. 제공하지 않으면 뷰포트의 스크린샷이 촬영됩니다. element가 제공된 경우 ref도 함께 제공해야 합니다.\n * `ref` (문자열, 선택 사항): 페이지 스냅샷의 정확한 대상 요소 참조입니다. 지정하지 않으면 뷰포트의 스크린샷이 생성됩니다. ref가 지정되면 요소도 지정해야 합니다.\n * 읽기 전용: **true**\n\n* **브라우저\\_PDF\\_저장**\n\n * 제목: PDF로 저장\n * 설명: 페이지를 PDF로 저장\n * 매개변수:\n * `filename` (문자열, 선택 사항): PDF를 저장할 파일 이름입니다. 지정하지 않으면 기본값은 `page-{timestamp}.pdf` 입니다.\n * 읽기 전용: **true**\n\n* **브라우저 네트워크 요청**\n\n * 제목: 네트워크 요청 목록\n * 설명: 페이지를 로드한 이후 모든 네트워크 요청을 반환합니다.\n * 매개변수: 없음\n * 읽기 전용: **true**\n\n* **브라우저 콘솔 메시지**\n\n * 제목: 콘솔 메시지 받기\n * 설명: 모든 콘솔 메시지를 반환합니다.\n * 매개변수: 없음\n * 읽기 전용: **true**\n\n* **브라우저 설치**\n\n * 제목 : config에 지정된 브라우저 설치\n * 설명: 설정에 지정된 브라우저를 설치합니다. 브라우저가 설치되지 않았다는 오류가 발생하면 이 메서드를 호출합니다.\n * 매개변수: 없음\n * 읽기 전용: **false**\n\n* **브라우저 닫기**\n\n * 제목: 브라우저 닫기\n * 설명: 페이지를 닫습니다\n * 매개변수: 없음\n * 읽기 전용: **true**\n\n* **브라우저 크기 조정**\n\n * 제목: 브라우저 창 크기 조정\n * 설명: 브라우저 창 크기 조정\n * 매개변수:\n * `width` (숫자): 브라우저 창의 너비\n * `height` (숫자): 브라우저 창의 높이\n * 읽기 전용: **true**\n\n* **브라우저 탭 목록**\n\n * 제목: 목록 탭\n * 설명: 브라우저 탭 나열\n * 매개변수: 없음\n * 읽기 전용: **true**\n\n* **브라우저\\_탭\\_새로 만들기**\n\n * 제목: 새 탭 열기\n * 설명: 새 탭을 엽니다\n * 매개변수:\n * `url` (문자열, 선택 사항): 새 탭에서 이동할 URL입니다. 지정하지 않으면 새 탭이 비어 있게 됩니다.\n * 읽기 전용: **true**\n\n* **브라우저\\_탭\\_선택**\n\n * 제목: 탭 선택\n * 설명: 인덱스로 탭 선택\n * 매개변수:\n * `index` (숫자): 선택할 탭의 인덱스\n * 읽기 전용: **true**\n\n* **브라우저\\_탭\\_닫기**\n\n * 제목: 탭 닫기\n * 설명: 탭 닫기\n * 매개변수:\n * `index` (숫자, 선택 사항): 닫을 탭의 인덱스입니다. 지정하지 않으면 현재 탭을 닫습니다.\n * 읽기 전용: **false**\n\n* **브라우저\\_생성\\_플레이라이트\\_테스트**\n\n * 제목: 극작가 테스트 생성\n * 설명: 주어진 시나리오에 대한 극작가 테스트 생성\n * 매개변수:\n * `name` (문자열): 테스트의 이름\n * `description` (문자열): 테스트에 대한 설명\n * `steps` (배열): 테스트의 단계\n * 읽기 전용: **true**\n\n* **브라우저 화면 캡처**\n\n * 제목: 스크린샷 찍기\n * 설명: 현재 페이지의 스크린샷을 찍습니다.\n * 매개변수: 없음\n * 읽기 전용: **true**\n\n* **브라우저 화면 마우스 이동**\n\n * 제목: 마우스 이동\n * 설명: 마우스를 주어진 위치로 이동합니다.\n * 매개변수:\n * `element` (문자열): 요소와 상호 작용할 수 있는 권한을 얻는 데 사용되는 사람이 읽을 수 있는 요소 설명\n * `x` (숫자): X 좌표\n * `y` (숫자): Y 좌표\n * 읽기 전용: **true**\n\n* **브라우저 화면 클릭**\n\n * 제목: 클릭\n * 설명: 마우스 왼쪽 버튼을 클릭하세요\n * 매개변수:\n * `element` (문자열): 요소와 상호 작용할 수 있는 권한을 얻는 데 사용되는 사람이 읽을 수 있는 요소 설명\n * `x` (숫자): X 좌표\n * `y` (숫자): Y 좌표\n * 읽기 전용: **false**\n\n* **브라우저 화면 드래그**\n\n * 제목: 드래그 마우스\n * 설명: 마우스 왼쪽 버튼 드래그\n * 매개변수:\n * `element` (문자열): 요소와 상호 작용할 수 있는 권한을 얻는 데 사용되는 사람이 읽을 수 있는 요소 설명\n * `startX` (숫자): 시작 X 좌표\n * `startY` (숫자): 시작 Y 좌표\n * `endX` (숫자): 끝 X 좌표\n * `endY` (숫자): 끝 Y 좌표\n * 읽기 전용: **false**\n\n* **브라우저\\_화면\\_유형**\n\n * 제목: 텍스트 입력\n * 설명: 텍스트를 입력하세요\n * 매개변수:\n * `text` (문자열): 요소에 입력할 텍스트\n * `submit` (부울, 선택 사항): 입력한 텍스트를 제출할지 여부(제출 후 Enter 키 누름)\n * 읽기 전용: **false**\n\n* **브라우저\\_누르기\\_키**\n\n * 제목: 키를 누르세요\n * 설명: 키보드의 키를 누르세요\n * 매개변수:\n * `key` (문자열): 누르려는 키의 이름 또는 생성할 문자(예: `ArrowLeft``a`\n * 읽기 전용: **false**\n\n* **브라우저 대기**\n\n * 제목 : 기다리다\n * 설명: 텍스트가 나타나거나 사라지거나 지정된 시간이 지날 때까지 기다립니다.\n * 매개변수:\n * `time` (숫자, 선택 사항): 대기 시간(초)\n * `text` (문자열, 선택 사항): 기다릴 텍스트\n * `textGone` (문자열, 선택 사항): 사라질 때까지 기다릴 텍스트\n * 읽기 전용: **true**\n\n* **브라우저 파일 업로드**\n\n * 제목: 파일 업로드\n * 설명: 하나 또는 여러 개의 파일을 업로드합니다.\n * 매개변수:\n * `paths` (배열): 업로드할 파일의 절대 경로입니다. 단일 파일 또는 여러 파일일 수 있습니다.\n * 읽기 전용: **false**\n\n* **브라우저\\_핸들\\_대화 상자**\n\n * 제목: 대화 처리\n * 설명: 대화 처리\n * 매개변수:\n * `accept` (boolean): 대화를 수락할지 여부.\n * `promptText` (문자열, 선택 사항): 프롬프트 대화 상자의 경우 프롬프트의 텍스트입니다.\n * 읽기 전용: **false**","Korean","ko-KR",{"_36":76,"_40":80,"_45":81},{"_38":77,"_40":78,"_42":79},"Сервер протокола контекста модели, который позволяет магистрам уровня магистра взаимодействовать с веб-страницами посредством структурированных снимков доступности без необходимости использования моделей зрения или снимков экрана.","Драматург MCP Сервер","## Драматург MCP\n\nСервер Model Context Protocol (MCP), который обеспечивает возможности автоматизации браузера с помощью [Playwright](https://playwright.dev) . Этот сервер позволяет LLM взаимодействовать с веб-страницами посредством структурированных снимков доступности, обходя необходимость в снимках экрана или визуально настроенных моделях.\n\n### Основные характеристики\n\n* **Быстрый и легкий** . Использует дерево доступности Playwright, а не пиксельный ввод.\n* **Дружелюбен к LLM** . Не требует моделей зрения, работает исключительно со структурированными данными.\n* **Детерминированное применение инструмента** . Избегает неоднозначности, характерной для подходов, основанных на скриншотах.\n\n### Требования\n\n* Node.js 18 или новее\n* VS Code, Cursor, Windsurf, Claude Desktop или любой другой MCP-клиент\n\n### Начиная\n\nСначала установите сервер Playwright MCP с вашим клиентом. Типичная конфигурация выглядит так:\n\n```js\n{\n \"mcpServers\": {\n \"playwright\": {\n \"command\": \"npx\",\n \"args\": [\n \"@playwright/mcp@latest\"\n ]\n }\n }\n}\n```\n\nВы также можете установить сервер Playwright MCP с помощью VS Code CLI:\n\n```bash\n# For VS Code\ncode --add-mcp '{\"name\":\"playwright\",\"command\":\"npx\",\"args\":[\"@playwright/mcp@latest\"]}'\n```\n\nПосле установки сервер Playwright MCP будет доступен для использования с вашим агентом GitHub Copilot в VS Code.\n\nПерейдите в `Cursor Settings` -> `MCP` -> `Add new MCP Server` . Назовите по своему вкусу, используйте тип `command` с командой `npx @playwright/mcp` . Вы также можете проверить конфигурацию или добавить аргументы, подобные команде, нажав `Edit` .\n\n```js\n{\n \"mcpServers\": {\n \"playwright\": {\n \"command\": \"npx\",\n \"args\": [\n \"@playwright/mcp@latest\"\n ]\n }\n }\n}\n```\n\nСледуйте [документации](https://docs.windsurf.com/windsurf/cascade/mcp) Windsuff MCP. Используйте следующую конфигурацию:\n\n```js\n{\n \"mcpServers\": {\n \"playwright\": {\n \"command\": \"npx\",\n \"args\": [\n \"@playwright/mcp@latest\"\n ]\n }\n }\n}\n```\n\nСледуя [руководству](https://modelcontextprotocol.io/quickstart/user) по установке MCP, используйте следующую конфигурацию:\n\n```js\n{\n \"mcpServers\": {\n \"playwright\": {\n \"command\": \"npx\",\n \"args\": [\n \"@playwright/mcp@latest\"\n ]\n }\n }\n}\n```\n\n### Конфигурация\n\nСервер Playwright MCP поддерживает следующие аргументы. Они могут быть предоставлены в конфигурации JSON выше, как часть списка `\"args\"` :\n\n```\n> npx @playwright/mcp@latest --help\n --allowed-origins semicolon-separated list of origins to allow the\n browser to request. Default is to allow all.\n --blocked-origins semicolon-separated list of origins to block the\n browser from requesting. Blocklist is evaluated\n before allowlist. If used without the allowlist,\n requests not matching the blocklist are still\n allowed.\n --block-service-workers block service workers\n --browser browser or chrome channel to use, possible\n values: chrome, firefox, webkit, msedge.\n --caps comma-separated list of capabilities to enable,\n possible values: tabs, pdf, history, wait, files,\n install. Default is all.\n --cdp-endpoint CDP endpoint to connect to.\n --config path to the configuration file.\n --device device to emulate, for example: \"iPhone 15\"\n --executable-path path to the browser executable.\n --headless run browser in headless mode, headed by default\n --host host to bind server to. Default is localhost. Use\n 0.0.0.0 to bind to all interfaces.\n --ignore-https-errors ignore https errors\n --isolated keep the browser profile in memory, do not save\n it to disk.\n --no-image-responses do not send image responses to the client.\n --no-sandbox disable the sandbox for all process types that\n are normally sandboxed.\n --output-dir path to the directory for output files.\n --port port to listen on for SSE transport.\n --proxy-bypass comma-separated domains to bypass proxy, for\n example \".com,chromium.org,.domain.com\"\n --proxy-server specify proxy server, for example\n \"http://myproxy:3128\" or \"socks5://myproxy:8080\"\n --save-trace Whether to save the Playwright Trace of the\n session into the output directory.\n --storage-state path to the storage state file for isolated\n sessions.\n --user-agent specify user agent string\n --user-data-dir path to the user data directory. If not\n specified, a temporary directory will be created.\n --viewport-size specify browser viewport size in pixels, for\n example \"1280, 720\"\n --vision Run server that uses screenshots (Aria snapshots\n are used by default)\n```\n\n### Профиль пользователя\n\nВы можете запустить Playwright MCP с постоянным профилем как обычный браузер (по умолчанию) или в изолированных контекстах для сеансов тестирования.\n\n**Постоянный профиль**\n\nВся информация о входе в систему будет храниться в постоянном профиле, вы можете удалить его между сеансами, если хотите очистить состояние офлайн. Постоянный профиль находится в следующих местах, и вы можете переопределить его с помощью аргумента `--user-data-dir` .\n\n```bash\n# Windows\n%USERPROFILE%\\AppData\\Local\\ms-playwright\\mcp-{channel}-profile\n\n# macOS\n- ~/Library/Caches/ms-playwright/mcp-{channel}-profile\n\n# Linux\n- ~/.cache/ms-playwright/mcp-{channel}-profile\n```\n\n**Изолированный**\n\nВ изолированном режиме каждый сеанс запускается в изолированном профиле. Каждый раз, когда вы просите MCP закрыть браузер, сеанс закрывается, и все состояние хранилища для этого сеанса теряется. Вы можете предоставить начальное состояние хранилища браузеру через `contextOptions` конфигурации или через аргумент `--storage-state` . Узнайте больше о состоянии хранилища [здесь](https://playwright.dev/docs/auth) .\n\n```js\n{\n \"mcpServers\": {\n \"playwright\": {\n \"command\": \"npx\",\n \"args\": [\n \"@playwright/mcp@latest\",\n \"--isolated\",\n \"--storage-state={path/to/storage.json}\n ]\n }\n }\n}\n```\n\n### Файл конфигурации\n\nСервер Playwright MCP можно настроить с помощью файла конфигурации JSON. Вы можете указать файл конфигурации с помощью параметра командной строки `--config` :\n\n```bash\nnpx @playwright/mcp@latest --config path/to/config.json\n```\n\n```typescript\n{\n // Browser configuration\n browser?: {\n // Browser type to use (chromium, firefox, or webkit)\n browserName?: 'chromium' | 'firefox' | 'webkit';\n\n // Keep the browser profile in memory, do not save it to disk.\n isolated?: boolean;\n\n // Path to user data directory for browser profile persistence\n userDataDir?: string;\n\n // Browser launch options (see Playwright docs)\n // @see https://playwright.dev/docs/api/class-browsertype#browser-type-launch\n launchOptions?: {\n channel?: string; // Browser channel (e.g. 'chrome')\n headless?: boolean; // Run in headless mode\n executablePath?: string; // Path to browser executable\n // ... other Playwright launch options\n };\n\n // Browser context options\n // @see https://playwright.dev/docs/api/class-browser#browser-new-context\n contextOptions?: {\n viewport?: { width: number, height: number };\n // ... other Playwright context options\n };\n\n // CDP endpoint for connecting to existing browser\n cdpEndpoint?: string;\n\n // Remote Playwright server endpoint\n remoteEndpoint?: string;\n },\n\n // Server configuration\n server?: {\n port?: number; // Port to listen on\n host?: string; // Host to bind to (default: localhost)\n },\n\n // List of enabled capabilities\n capabilities?: Array<\n 'core' | // Core browser automation\n 'tabs' | // Tab management\n 'pdf' | // PDF generation\n 'history' | // Browser history\n 'wait' | // Wait utilities\n 'files' | // File handling\n 'install' | // Browser installation\n 'testing' // Testing\n >;\n\n // Enable vision mode (screenshots instead of accessibility snapshots)\n vision?: boolean;\n\n // Directory for output files\n outputDir?: string;\n\n // Network configuration\n network?: {\n // List of origins to allow the browser to request. Default is to allow all. Origins matching both `allowedOrigins` and `blockedOrigins` will be blocked.\n allowedOrigins?: string[];\n\n // List of origins to block the browser to request. Origins matching both `allowedOrigins` and `blockedOrigins` will be blocked.\n blockedOrigins?: string[];\n };\n \n /**\n * Do not send image responses to the client.\n */\n noImageResponses?: boolean;\n}\n```\n\n### Автономный MCP-сервер\n\nПри запуске браузера Headed в системе без дисплея или из рабочих процессов IDE запустите сервер MCP из среды с DISPLAY и передайте флаг `--port` , чтобы включить транспорт SSE.\n\n```bash\nnpx @playwright/mcp@latest --port 8931\n```\n\nЗатем в конфигурации клиента MCP задайте `url` -адрес конечной точки SSE:\n\n```js\n{\n \"mcpServers\": {\n \"playwright\": {\n \"url\": \"http://localhost:8931/sse\"\n }\n }\n}\n```\n\n**ПРИМЕЧАНИЕ:** На данный момент реализация Docker поддерживает только безголовый Chromium.\n\n```js\n{\n \"mcpServers\": {\n \"playwright\": {\n \"command\": \"docker\",\n \"args\": [\"run\", \"-i\", \"--rm\", \"--init\", \"--pull=always\", \"mcr.microsoft.com/playwright/mcp\"]\n }\n }\n}\n```\n\nВы можете создать образ Docker самостоятельно.\n\n```\ndocker build -t mcr.microsoft.com/playwright/mcp .\n```\n\n```js\nimport http from 'http';\n\nimport { createConnection } from '@playwright/mcp';\nimport { SSEServerTransport } from '@modelcontextprotocol/sdk/server/sse.js';\n\nhttp.createServer(async (req, res) => {\n // ...\n\n // Creates a headless Playwright MCP server with SSE transport\n const connection = await createConnection({ browser: { launchOptions: { headless: true } } });\n const transport = new SSEServerTransport('/messages', res);\n await connection.connect(transport);\n\n // ...\n});\n```\n\n### Инструменты\n\nИнструменты доступны в двух режимах:\n\n1. **Режим моментальных снимков** (по умолчанию): использует моментальные снимки доступности для повышения производительности и надежности.\n2. **Режим Vision** : использует снимки экрана для визуального взаимодействия.\n\nЧтобы использовать Vision Mode, добавьте флаг `--vision` при запуске сервера:\n\n```js\n{\n \"mcpServers\": {\n \"playwright\": {\n \"command\": \"npx\",\n \"args\": [\n \"@playwright/mcp@latest\",\n \"--vision\"\n ]\n }\n }\n}\n```\n\nРежим Vision лучше всего работает с компьютерными моделями, которые способны взаимодействовать с элементами, используя координатное пространство XY, судя по предоставленному снимку экрана.\n\n* **снимок\\_браузера**\n\n * Название: Снимок страницы\n * Описание: Сделать снимок доступности текущей страницы, это лучше, чем снимок экрана\n * Параметры: Нет\n * Только для чтения: **правда**\n\n* **браузер\\_щелчок**\n\n * Название: Щелчок\n * Описание: Выполнить щелчок на веб-странице.\n * Параметры:\n * `element` (строка): понятное человеку описание элемента, используемое для получения разрешения на взаимодействие с элементом\n * `ref` (string): Точная ссылка на целевой элемент из снимка страницы\n * Только для чтения: **ложно**\n\n* **браузер\\_перетаскивание**\n\n * Название: Перетаскивание мыши\n * Описание: Выполнение перетаскивания между двумя элементами.\n * Параметры:\n * `startElement` (строка): понятное человеку описание исходного элемента, используемое для получения разрешения на взаимодействие с элементом.\n * `startRef` (string): Точная ссылка на исходный элемент из снимка страницы\n * `endElement` (строка): понятное человеку описание целевого элемента, используемое для получения разрешения на взаимодействие с элементом.\n * `endRef` (строка): точная ссылка на целевой элемент из снимка страницы\n * Только для чтения: **ложно**\n\n* **браузер\\_hover**\n\n * Название: Наведение мыши\n * Описание: Наведите курсор на элемент на странице.\n * Параметры:\n * `element` (строка): понятное человеку описание элемента, используемое для получения разрешения на взаимодействие с элементом\n * `ref` (string): Точная ссылка на целевой элемент из снимка страницы\n * Только для чтения: **правда**\n\n* **тип\\_браузера**\n\n * Название: Введите текст\n * Описание: Введите текст в редактируемый элемент.\n * Параметры:\n * `element` (строка): понятное человеку описание элемента, используемое для получения разрешения на взаимодействие с элементом\n * `ref` (string): Точная ссылка на целевой элемент из снимка страницы\n * `text` (строка): Текст для ввода в элемент.\n * `submit` (логическое значение, необязательно): следует ли отправить введенный текст (нажмите Enter после этого)\n * `slowly` (boolean, необязательно): Вводить ли по одному символу за раз. Полезно для запуска обработчиков клавиш на странице. По умолчанию весь текст заполняется сразу.\n * Только для чтения: **ложно**\n\n* **выбор\\_опции\\_браузера**\n\n * Название: Выберите вариант\n * Описание: Выберите вариант в раскрывающемся списке.\n * Параметры:\n * `element` (строка): понятное человеку описание элемента, используемое для получения разрешения на взаимодействие с элементом\n * `ref` (string): Точная ссылка на целевой элемент из снимка страницы\n * `values` (массив): Массив значений для выбора в раскрывающемся списке. Это может быть одно значение или несколько значений.\n * Только для чтения: **ложно**\n\n* **браузер\\_нажмите\\_клавишу**\n\n * Название: Нажмите клавишу\n * Описание: Нажмите клавишу на клавиатуре.\n * Параметры:\n * `key` (строка): название клавиши, которую нужно нажать, или символ, который нужно сгенерировать, например, `ArrowLeft``a`\n * Только для чтения: **ложно**\n\n* **браузер\\_ожидание\\_для**\n\n * Название: Подождите\n * Описание: Ожидание появления или исчезновения текста или истечения указанного времени.\n * Параметры:\n * `time` (число, необязательно): время ожидания в секундах\n * `text` (строка, необязательно): Текст, которого нужно ждать\n * `textGone` (строка, необязательно): Текст, которого нужно дождаться, чтобы исчезнуть.\n * Только для чтения: **правда**\n\n* **загрузка\\_файла\\_браузера**\n\n * Название: Загрузить файлы\n * Описание: Загрузите один или несколько файлов.\n * Параметры:\n * `paths` (массив): Абсолютные пути к файлам для загрузки. Может быть одним файлом или несколькими файлами.\n * Только для чтения: **ложно**\n\n* **диалог\\_обработки\\_браузера**\n\n * Название: Обработка диалога\n * Описание: Обработка диалога.\n * Параметры:\n * `accept` (логическое значение): принимать ли диалог.\n * `promptText` (строка, необязательно): Текст подсказки в случае диалогового окна подсказки.\n * Только для чтения: **ложно**\n\n* **браузер\\_навигация**\n\n * Название: Перейти к URL-адресу\n * Описание: Перейдите по URL-адресу\n * Параметры:\n * `url` (строка): URL-адрес для перехода\n * Только для чтения: **ложно**\n\n* **браузер\\_навигация\\_назад**\n\n * Название: Вернуться назад\n * Описание: Вернуться на предыдущую страницу\n * Параметры: Нет\n * Только для чтения: **правда**\n\n* **браузер\\_навигация\\_вперед**\n\n * Название: Идти вперед\n * Описание: Перейти на следующую страницу\n * Параметры: Нет\n * Только для чтения: **правда**\n\n* **браузер\\_сделать\\_скриншот**\n\n * Название: Сделать снимок экрана\n * Описание: Сделать снимок экрана текущей страницы. Вы не можете выполнять действия на основе снимка экрана, используйте для действий browser\\_snapshot.\n * Параметры:\n * `raw` (boolean, необязательно): Возвращать ли без сжатия (в формате PNG). Значение по умолчанию — false, что возвращает изображение JPEG.\n * `filename` (строка, необязательно): Имя файла для сохранения снимка экрана. По умолчанию `page-{timestamp}.{png|jpeg}` если не указано иное.\n * `element` (строка, необязательно): понятное человеку описание элемента, используемое для получения разрешения на снимок экрана элемента. Если не указано, снимок экрана будет сделан из области просмотра. Если элемент указан, необходимо указать также ссылку.\n * `ref` (строка, необязательно): Точная ссылка на целевой элемент из снимка страницы. Если не указано, снимок экрана будет сделан из области просмотра. Если указано ref, элемент также должен быть указан.\n * Только для чтения: **правда**\n\n* **браузер\\_pdf\\_сохранить**\n\n * Название: Сохранить как PDF\n * Описание: Сохранить страницу как PDF\n * Параметры:\n * `filename` (строка, необязательно): Имя файла для сохранения PDF-файла. По умолчанию `page-{timestamp}.pdf` если не указано иное.\n * Только для чтения: **правда**\n\n* **запросы\\_браузера\\_сети**\n\n * Название: Список сетевых запросов\n * Описание: Возвращает все сетевые запросы с момента загрузки страницы.\n * Параметры: Нет\n * Только для чтения: **правда**\n\n* **сообщения\\_консоли\\_браузера**\n\n * Название: Получить консольные сообщения\n * Описание: Возвращает все сообщения консоли.\n * Параметры: Нет\n * Только для чтения: **правда**\n\n* **браузер\\_установить**\n\n * Заголовок: Установка браузера, указанного в конфиге\n * Описание: Установите браузер, указанный в конфигурации. Вызовите его, если вы получили ошибку о том, что браузер не установлен.\n * Параметры: Нет\n * Только для чтения: **ложно**\n\n* **браузер\\_закрыть**\n\n * Заголовок: Закрыть браузер\n * Описание: Закрыть страницу\n * Параметры: Нет\n * Только для чтения: **правда**\n\n* **изменение\\_размера\\_браузера**\n\n * Название: Изменение размера окна браузера\n * Описание: Изменение размера окна браузера.\n * Параметры:\n * `width` (число): Ширина окна браузера.\n * `height` (число): Высота окна браузера.\n * Только для чтения: **правда**\n\n* **список\\_вкладок\\_браузера**\n\n * Название: Список вкладок\n * Описание: Список вкладок браузера\n * Параметры: Нет\n * Только для чтения: **правда**\n\n* **браузер\\_вкладка\\_новая**\n\n * Заголовок: Открыть новую вкладку\n * Описание: Открыть новую вкладку\n * Параметры:\n * `url` (строка, необязательно): URL для перехода на новую вкладку. Если не указано, новая вкладка будет пустой.\n * Только для чтения: **правда**\n\n* **выбор\\_вкладки\\_браузера**\n\n * Название: Выберите вкладку\n * Описание: Выберите вкладку по индексу\n * Параметры:\n * `index` (номер): индекс вкладки для выбора.\n * Только для чтения: **правда**\n\n* **закрыть\\_вкладку\\_браузера**\n\n * Заголовок: Закрыть вкладку\n * Описание: Закрыть вкладку\n * Параметры:\n * `index` (число, необязательно): индекс вкладки для закрытия. Закрывает текущую вкладку, если не указано.\n * Только для чтения: **ложно**\n\n* **браузер\\_генерировать\\_драматург\\_тест**\n\n * Название: Сгенерируйте тест драматурга\n * Описание: Создать тест драматурга для заданного сценария.\n * Параметры:\n * `name` (строка): Название теста\n * `description` (строка): Описание теста\n * `steps` (массив): шаги теста\n * Только для чтения: **правда**\n\n* **захват\\_экрана\\_браузера**\n\n * Название: Сделать снимок экрана\n * Описание: Сделать снимок экрана текущей страницы.\n * Параметры: Нет\n * Только для чтения: **правда**\n\n* **экран\\_браузера\\_перемещение\\_мыши**\n\n * Название: Перемещение мыши\n * Описание: Переместить мышь в указанную позицию.\n * Параметры:\n * `element` (строка): понятное человеку описание элемента, используемое для получения разрешения на взаимодействие с элементом\n * `x` (число): координата X\n * `y` (число): координата Y\n * Только для чтения: **правда**\n\n* **щелчок\\_по\\_экрану\\_браузера**\n\n * Название: Щелчок\n * Описание: Щелкните левой кнопкой мыши.\n * Параметры:\n * `element` (строка): понятное человеку описание элемента, используемое для получения разрешения на взаимодействие с элементом\n * `x` (число): координата X\n * `y` (число): координата Y\n * Только для чтения: **ложно**\n\n* **перетаскивание\\_экрана\\_браузера**\n\n * Название: Перетаскивание мыши\n * Описание: Перетащите левой кнопкой мыши\n * Параметры:\n * `element` (строка): понятное человеку описание элемента, используемое для получения разрешения на взаимодействие с элементом\n * `startX` (число): Начальная координата X\n * `startY` (число): начальная координата Y\n * `endX` (число): Конечная координата X\n * `endY` (число): Конечная координата Y\n * Только для чтения: **ложно**\n\n* **тип\\_экрана\\_браузера**\n\n * Название: Введите текст\n * Описание: Введите текст\n * Параметры:\n * `text` (строка): Текст для ввода в элемент.\n * `submit` (логическое значение, необязательно): следует ли отправить введенный текст (нажмите Enter после этого)\n * Только для чтения: **ложно**\n\n* **браузер\\_нажмите\\_клавишу**\n\n * Название: Нажмите клавишу\n * Описание: Нажмите клавишу на клавиатуре.\n * Параметры:\n * `key` (строка): название клавиши, которую нужно нажать, или символ, который нужно сгенерировать, например, `ArrowLeft``a`\n * Только для чтения: **ложно**\n\n* **браузер\\_ожидание\\_для**\n\n * Название: Подождите\n * Описание: Ожидание появления или исчезновения текста или истечения указанного времени.\n * Параметры:\n * `time` (число, необязательно): время ожидания в секундах\n * `text` (строка, необязательно): Текст, которого нужно ждать\n * `textGone` (строка, необязательно): Текст, которого нужно дождаться, чтобы исчезнуть.\n * Только для чтения: **правда**\n\n* **загрузка\\_файла\\_браузера**\n\n * Название: Загрузить файлы\n * Описание: Загрузите один или несколько файлов.\n * Параметры:\n * `paths` (массив): Абсолютные пути к файлам для загрузки. Может быть одним файлом или несколькими файлами.\n * Только для чтения: **ложно**\n\n* **диалог\\_обработки\\_браузера**\n\n * Название: Обработка диалога\n * Описание: Обработка диалога.\n * Параметры:\n * `accept` (логическое значение): принимать ли диалог.\n * `promptText` (строка, необязательно): Текст подсказки в случае диалогового окна подсказки.\n * Только для чтения: **ложно**","Russian","ru-RU",{"_36":83,"_40":87,"_45":88},{"_38":84,"_40":85,"_42":86},"模型上下文协议服务器使 LLM 能够通过结构化的可访问性快照与网页交互，而无需视觉模型或屏幕截图。","剧作家 MCP 服务器","## 剧作家 MCP\n\n一个使用[Playwright](https://playwright.dev)提供浏览器自动化功能的模型上下文协议 (MCP) 服务器。该服务器使 LLM 能够通过结构化的可访问性快照与网页进行交互，而无需使用屏幕截图或视觉调整的模型。\n\n### 主要特点\n\n* **快速且轻量**。使用 Playwright 的可访问性树，而不是基于像素的输入。\n* **适合法学硕士 (LLM)** 。无需视觉模型，纯粹基于结构化数据运行。\n* **确定性工具应用程序**。避免了基于屏幕截图的方法中常见的歧义。\n\n### 要求\n\n* Node.js 18 或更高版本\n* VS Code、Cursor、Windsurf、Claude Desktop 或任何其他 MCP 客户端\n\n### 入门\n\n首先，在您的客户端上安装 Playwright MCP 服务器。典型的配置如下：\n\n```js\n{\n \"mcpServers\": {\n \"playwright\": {\n \"command\": \"npx\",\n \"args\": [\n \"@playwright/mcp@latest\"\n ]\n }\n }\n}\n```\n\n您还可以使用 VS Code CLI 安装 Playwright MCP 服务器：\n\n```bash\n# For VS Code\ncode --add-mcp '{\"name\":\"playwright\",\"command\":\"npx\",\"args\":[\"@playwright/mcp@latest\"]}'\n```\n\n安装后，Playwright MCP 服务器将可与 VS Code 中的 GitHub Copilot 代理一起使用。\n\n前往`Cursor Settings` -> `MCP` -> `Add new MCP Server` 。根据个人喜好命名，使用`command`类型`npx @playwright/mcp` 。您也可以点击`Edit`来验证配置或添加命令参数。\n\n```js\n{\n \"mcpServers\": {\n \"playwright\": {\n \"command\": \"npx\",\n \"args\": [\n \"@playwright/mcp@latest\"\n ]\n }\n }\n}\n```\n\n按照 Windsuff MCP[文档](https://docs.windsurf.com/windsurf/cascade/mcp)操作。使用以下配置：\n\n```js\n{\n \"mcpServers\": {\n \"playwright\": {\n \"command\": \"npx\",\n \"args\": [\n \"@playwright/mcp@latest\"\n ]\n }\n }\n}\n```\n\n按照 MCP 安装[指南](https://modelcontextprotocol.io/quickstart/user)，使用以下配置：\n\n```js\n{\n \"mcpServers\": {\n \"playwright\": {\n \"command\": \"npx\",\n \"args\": [\n \"@playwright/mcp@latest\"\n ]\n }\n }\n}\n```\n\n### 配置\n\nPlaywright MCP 服务器支持以下参数。它们可以在上面的 JSON 配置中作为`\"args\"`列表的一部分提供：\n\n```\n> npx @playwright/mcp@latest --help\n --allowed-origins semicolon-separated list of origins to allow the\n browser to request. Default is to allow all.\n --blocked-origins semicolon-separated list of origins to block the\n browser from requesting. Blocklist is evaluated\n before allowlist. If used without the allowlist,\n requests not matching the blocklist are still\n allowed.\n --block-service-workers block service workers\n --browser browser or chrome channel to use, possible\n values: chrome, firefox, webkit, msedge.\n --caps comma-separated list of capabilities to enable,\n possible values: tabs, pdf, history, wait, files,\n install. Default is all.\n --cdp-endpoint CDP endpoint to connect to.\n --config path to the configuration file.\n --device device to emulate, for example: \"iPhone 15\"\n --executable-path path to the browser executable.\n --headless run browser in headless mode, headed by default\n --host host to bind server to. Default is localhost. Use\n 0.0.0.0 to bind to all interfaces.\n --ignore-https-errors ignore https errors\n --isolated keep the browser profile in memory, do not save\n it to disk.\n --no-image-responses do not send image responses to the client.\n --no-sandbox disable the sandbox for all process types that\n are normally sandboxed.\n --output-dir path to the directory for output files.\n --port port to listen on for SSE transport.\n --proxy-bypass comma-separated domains to bypass proxy, for\n example \".com,chromium.org,.domain.com\"\n --proxy-server specify proxy server, for example\n \"http://myproxy:3128\" or \"socks5://myproxy:8080\"\n --save-trace Whether to save the Playwright Trace of the\n session into the output directory.\n --storage-state path to the storage state file for isolated\n sessions.\n --user-agent specify user agent string\n --user-data-dir path to the user data directory. If not\n specified, a temporary directory will be created.\n --viewport-size specify browser viewport size in pixels, for\n example \"1280, 720\"\n --vision Run server that uses screenshots (Aria snapshots\n are used by default)\n```\n\n### 用户个人资料\n\n您可以像常规浏览器（默认）一样使用持久配置文件运行 Playwright MCP，或者在测试会话的隔离环境中运行。\n\n**持久配置文件**\n\n所有登录信息都将存储在持久配置文件中，如果您想清除离线状态，可以在会话之间删除它。持久配置文件位于以下位置，您可以使用`--user-data-dir`参数覆盖它。\n\n```bash\n# Windows\n%USERPROFILE%\\AppData\\Local\\ms-playwright\\mcp-{channel}-profile\n\n# macOS\n- ~/Library/Caches/ms-playwright/mcp-{channel}-profile\n\n# Linux\n- ~/.cache/ms-playwright/mcp-{channel}-profile\n```\n\n**孤立**\n\n在隔离模式下，每个会话都以隔离配置文件启动。每次您要求 MCP 关闭浏览器时，会话都会关闭，并且此会话的所有存储状态都会丢失。您可以通过配置的`contextOptions`或`--storage-state`参数向浏览器提供初始存储状态。点击[此处](https://playwright.dev/docs/auth)了解更多关于存储状态的信息。\n\n```js\n{\n \"mcpServers\": {\n \"playwright\": {\n \"command\": \"npx\",\n \"args\": [\n \"@playwright/mcp@latest\",\n \"--isolated\",\n \"--storage-state={path/to/storage.json}\n ]\n }\n }\n}\n```\n\n### 配置文件\n\nPlaywright MCP 服务器可以使用 JSON 配置文件进行配置。您可以使用`--config`命令行选项指定配置文件：\n\n```bash\nnpx @playwright/mcp@latest --config path/to/config.json\n```\n\n```typescript\n{\n // Browser configuration\n browser?: {\n // Browser type to use (chromium, firefox, or webkit)\n browserName?: 'chromium' | 'firefox' | 'webkit';\n\n // Keep the browser profile in memory, do not save it to disk.\n isolated?: boolean;\n\n // Path to user data directory for browser profile persistence\n userDataDir?: string;\n\n // Browser launch options (see Playwright docs)\n // @see https://playwright.dev/docs/api/class-browsertype#browser-type-launch\n launchOptions?: {\n channel?: string; // Browser channel (e.g. 'chrome')\n headless?: boolean; // Run in headless mode\n executablePath?: string; // Path to browser executable\n // ... other Playwright launch options\n };\n\n // Browser context options\n // @see https://playwright.dev/docs/api/class-browser#browser-new-context\n contextOptions?: {\n viewport?: { width: number, height: number };\n // ... other Playwright context options\n };\n\n // CDP endpoint for connecting to existing browser\n cdpEndpoint?: string;\n\n // Remote Playwright server endpoint\n remoteEndpoint?: string;\n },\n\n // Server configuration\n server?: {\n port?: number; // Port to listen on\n host?: string; // Host to bind to (default: localhost)\n },\n\n // List of enabled capabilities\n capabilities?: Array<\n 'core' | // Core browser automation\n 'tabs' | // Tab management\n 'pdf' | // PDF generation\n 'history' | // Browser history\n 'wait' | // Wait utilities\n 'files' | // File handling\n 'install' | // Browser installation\n 'testing' // Testing\n >;\n\n // Enable vision mode (screenshots instead of accessibility snapshots)\n vision?: boolean;\n\n // Directory for output files\n outputDir?: string;\n\n // Network configuration\n network?: {\n // List of origins to allow the browser to request. Default is to allow all. Origins matching both `allowedOrigins` and `blockedOrigins` will be blocked.\n allowedOrigins?: string[];\n\n // List of origins to block the browser to request. Origins matching both `allowedOrigins` and `blockedOrigins` will be blocked.\n blockedOrigins?: string[];\n };\n \n /**\n * Do not send image responses to the client.\n */\n noImageResponses?: boolean;\n}\n```\n\n### 独立 MCP 服务器\n\n在没有显示器的系统上或从 IDE 的工作进程运行带头浏览器时，从具有 DISPLAY 的环境中运行 MCP 服务器并传递`--port`标志以启用 SSE 传输。\n\n```bash\nnpx @playwright/mcp@latest --port 8931\n```\n\n然后在 MCP 客户端配置中，将`url`设置为 SSE 端点：\n\n```js\n{\n \"mcpServers\": {\n \"playwright\": {\n \"url\": \"http://localhost:8931/sse\"\n }\n }\n}\n```\n\n**注意：** Docker 实现目前仅支持无头 Chromium。\n\n```js\n{\n \"mcpServers\": {\n \"playwright\": {\n \"command\": \"docker\",\n \"args\": [\"run\", \"-i\", \"--rm\", \"--init\", \"--pull=always\", \"mcr.microsoft.com/playwright/mcp\"]\n }\n }\n}\n```\n\n您可以自己构建 Docker 镜像。\n\n```\ndocker build -t mcr.microsoft.com/playwright/mcp .\n```\n\n```js\nimport http from 'http';\n\nimport { createConnection } from '@playwright/mcp';\nimport { SSEServerTransport } from '@modelcontextprotocol/sdk/server/sse.js';\n\nhttp.createServer(async (req, res) => {\n // ...\n\n // Creates a headless Playwright MCP server with SSE transport\n const connection = await createConnection({ browser: { launchOptions: { headless: true } } });\n const transport = new SSEServerTransport('/messages', res);\n await connection.connect(transport);\n\n // ...\n});\n```\n\n### 工具\n\n该工具有两种模式：\n\n1. **快照模式**（默认）：使用可访问性快照来获得更好的性能和可靠性\n2. **视觉模式**：使用屏幕截图进行基于视觉的交互\n\n要使用 Vision 模式，请在启动服务器时添加`--vision`标志：\n\n```js\n{\n \"mcpServers\": {\n \"playwright\": {\n \"command\": \"npx\",\n \"args\": [\n \"@playwright/mcp@latest\",\n \"--vision\"\n ]\n }\n }\n}\n```\n\n根据提供的屏幕截图，视觉模式最适合能够使用 XY 坐标空间与元素交互的计算机使用模型。\n\n* **browser\\_snapshot**\n\n * 标题：页面快照\n * 描述：捕获当前页面的可访问性快照，这比屏幕截图更好\n * 参数：无\n * 只读： **true**\n\n* **browser\\_click**\n\n * 标题：点击\n * 描述：执行网页上的点击操作\n * 参数：\n * `element` （字符串）：用于获取与元素交互的权限的人类可读的元素描述\n * `ref` （字符串）：来自页面快照的精确目标元素引用\n * 只读： **false**\n\n* **browser\\_drag**\n\n * 标题：拖动鼠标\n * 描述：在两个元素之间执行拖放操作\n * 参数：\n * `startElement` （字符串）：用于获取与元素交互权限的人类可读的源元素描述\n * `startRef` （字符串）：来自页面快照的精确源元素引用\n * `endElement` （字符串）：用于获取与元素交互的权限的人类可读的目标元素描述\n * `endRef` （字符串）：来自页面快照的精确目标元素引用\n * 只读： **false**\n\n* **browser\\_hover**\n\n * 标题：鼠标悬停\n * 描述：将鼠标悬停在页面上的元素上\n * 参数：\n * `element` （字符串）：用于获取与元素交互的权限的人类可读的元素描述\n * `ref` （字符串）：来自页面快照的精确目标元素引用\n * 只读： **true**\n\n* **浏览器类型**\n\n * 标题：输入文本\n * 描述：在可编辑元素中输入文本\n * 参数：\n * `element` （字符串）：用于获取与元素交互的权限的人类可读的元素描述\n * `ref` （字符串）：来自页面快照的精确目标元素引用\n * `text` （字符串）：输入到元素中的文本\n * `submit` （布尔值，可选）：是否提交输入的文本（按 Enter 键后）\n * `slowly` （布尔值，可选）：是否一次输入一个字符。用于触发页面中的按键处理程序。默认情况下，整个文本会一次性填充。\n * 只读： **false**\n\n* **browser\\_select\\_option**\n\n * 标题：选择选项\n * 描述：在下拉菜单中选择一个选项\n * 参数：\n * `element` （字符串）：用于获取与元素交互的权限的人类可读的元素描述\n * `ref` （字符串）：来自页面快照的精确目标元素引用\n * `values` （数组）：下拉菜单中要选择的值的数组。可以是单个值，也可以是多个值。\n * 只读： **false**\n\n* **browser\\_press\\_key**\n\n * 标题：按下一个键\n * 描述：按下键盘上的一个键\n * 参数：\n * `key` （字符串）：要按下的键的名称或要生成的字符，例如`ArrowLeft``a`\n * 只读： **false**\n\n* **browser\\_wait\\_for**\n\n * 标题：等待\n * 描述：等待文本出现或消失或等待指定的时间过去\n * 参数：\n * `time` （数字，可选）：等待的时间（秒）\n * `text` （字符串，可选）：等待的文本\n * `textGone` （字符串，可选）：等待消失的文本\n * 只读： **true**\n\n* **browser\\_file\\_upload**\n\n * 标题：上传文件\n * 描述：上传一个或多个文件\n * 参数：\n * `paths` （数组）：要上传的文件的绝对路径。可以是单个文件，也可以是多个文件。\n * 只读： **false**\n\n* **browser\\_handle\\_dialog**\n\n * 标题：处理对话\n * 描述：处理对话\n * 参数：\n * `accept` （布尔值）：是否接受对话。\n * `promptText` （字符串，可选）：提示对话框中的提示文本。\n * 只读： **false**\n\n* **browser\\_navigate**\n\n * 标题：导航至 URL\n * 描述：导航至 URL\n * 参数：\n * `url` （字符串）：要导航到的 URL\n * 只读： **false**\n\n* **browser\\_navigate\\_back**\n\n * 标题：回去\n * 描述：返回上一页\n * 参数：无\n * 只读： **true**\n\n* **browser\\_navigate\\_forward**\n\n * 标题：前进\n * 描述：前进到下一页\n * 参数：无\n * 只读： **true**\n\n* **browser\\_take\\_screenshot**\n\n * 标题：截取屏幕截图\n * 描述：截取当前页面的屏幕截图。您无法基于屏幕截图执行任何操作，请使用 browser\\_snapshot 执行操作。\n * 参数：\n * `raw` （布尔值，可选）：是否返回未压缩的图像（PNG 格式）。默认值为 false，即返回 JPEG 格式的图像。\n * `filename` （字符串，可选）：保存屏幕截图的文件名。如果未指定，则默认为`page-{timestamp}.{png|jpeg}` 。\n * `element` （字符串，可选）：用于获取元素截图权限的可读元素描述。若未提供，则截取视口范围的屏幕截图。若提供 element，则必须同时提供 ref。\n * `ref` （字符串，可选）：页面快照中目标元素的精确引用。若未提供，则截取视口的屏幕截图。若提供 ref，则必须同时提供元素。\n * 只读： **true**\n\n* **browser\\_pdf\\_save**\n\n * 标题：另存为 PDF\n * 描述：将页面保存为 PDF\n * 参数：\n * `filename` （字符串，可选）：保存 PDF 文件的文件名。若未指定，则默认为`page-{timestamp}.pdf` 。\n * 只读： **true**\n\n* **browser\\_network\\_requests**\n\n * 标题：列出网络请求\n * 描述：返回自加载页面以来的所有网络请求\n * 参数：无\n * 只读： **true**\n\n* **browser\\_console\\_messages**\n\n * 标题：获取控制台消息\n * 描述：返回所有控制台消息\n * 参数：无\n * 只读： **true**\n\n* **浏览器安装**\n\n * 标题：安装配置中指定的浏览器\n * 描述：安装配置中指定的浏览器。如果出现浏览器未安装的错误，请调用此函数。\n * 参数：无\n * 只读： **false**\n\n* **browser\\_close**\n\n * 标题：关闭浏览器\n * 描述：关闭页面\n * 参数：无\n * 只读： **true**\n\n* **browser\\_resize**\n\n * 标题：调整浏览器窗口大小\n * 描述：调整浏览器窗口的大小\n * 参数：\n * `width` （数字）：浏览器窗口的宽度\n * `height` （数字）：浏览器窗口的高度\n * 只读： **true**\n\n* **browser\\_tab\\_list**\n\n * 标题：列表选项卡\n * 描述：列出浏览器标签页\n * 参数：无\n * 只读： **true**\n\n* **browser\\_tab\\_new**\n\n * 标题：打开新标签页\n * 描述：打开新标签页\n * 参数：\n * `url` （字符串，可选）：新标签页中导航到的 URL。如果不提供，新标签页将为空白。\n * 只读： **true**\n\n* **browser\\_tab\\_select**\n\n * 标题：选择一个选项卡\n * 描述：通过索引选择选项卡\n * 参数：\n * `index` （数字）：要选择的选项卡的索引\n * 只读： **true**\n\n* **browser\\_tab\\_close**\n\n * 标题：关闭标签页\n * 描述：关闭标签页\n * 参数：\n * `index` （数字，可选）：要关闭的标签页的索引。如果未提供，则关闭当前标签页。\n * 只读： **false**\n\n* **browser\\_generate\\_playwright\\_test**\n\n * 标题：生成 Playwright 测试\n * 描述：为给定场景生成 Playwright 测试\n * 参数：\n * `name` （字符串）：测试的名称\n * `description` （字符串）：测试的描述\n * `steps` （数组）：测试的步骤\n * 只读： **true**\n\n* **browser\\_screen\\_capture**\n\n * 标题：截取屏幕截图\n * 描述：截取当前页面的屏幕截图\n * 参数：无\n * 只读： **true**\n\n* **browser\\_screen\\_move\\_mouse**\n\n * 标题：移动鼠标\n * 描述：移动鼠标到指定位置\n * 参数：\n * `element` （字符串）：用于获取与元素交互的权限的人类可读的元素描述\n * `x` （数字）：X 坐标\n * `y` （数字）：Y 坐标\n * 只读： **true**\n\n* **browser\\_screen\\_click**\n\n * 标题：点击\n * 描述：单击鼠标左键\n * 参数：\n * `element` （字符串）：用于获取与元素交互的权限的人类可读的元素描述\n * `x` （数字）：X 坐标\n * `y` （数字）：Y 坐标\n * 只读： **false**\n\n* **browser\\_screen\\_drag**\n\n * 标题：拖动鼠标\n * 描述：拖动鼠标左键\n * 参数：\n * `element` （字符串）：用于获取与元素交互的权限的人类可读的元素描述\n * `startX` （数字）：起始X坐标\n * `startY` （数字）：起始 Y 坐标\n * `endX` （数字）：结束X坐标\n * `endY` （数字）：结束 Y 坐标\n * 只读： **false**\n\n* **浏览器屏幕类型**\n\n * 标题：输入文本\n * 描述：输入文本\n * 参数：\n * `text` （字符串）：输入到元素中的文本\n * `submit` （布尔值，可选）：是否提交输入的文本（按 Enter 键后）\n * 只读： **false**\n\n* **browser\\_press\\_key**\n\n * 标题：按下一个键\n * 描述：按下键盘上的一个键\n * 参数：\n * `key` （字符串）：要按下的键的名称或要生成的字符，例如`ArrowLeft``a`\n * 只读： **false**\n\n* **browser\\_wait\\_for**\n\n * 标题：等待\n * 描述：等待文本出现或消失或等待指定的时间过去\n * 参数：\n * `time` （数字，可选）：等待的时间（秒）\n * `text` （字符串，可选）：等待的文本\n * `textGone` （字符串，可选）：等待消失的文本\n * 只读： **true**\n\n* **browser\\_file\\_upload**\n\n * 标题：上传文件\n * 描述：上传一个或多个文件\n * 参数：\n * `paths` （数组）：要上传的文件的绝对路径。可以是单个文件，也可以是多个文件。\n * 只读： **false**\n\n* **browser\\_handle\\_dialog**\n\n * 标题：处理对话\n * 描述：处理对话\n * 参数：\n * `accept` （布尔值）：是否接受对话。\n * `promptText` （字符串，可选）：提示对话框中的提示文本。\n * 只读： **false**","Chinese","zh-CN","githubUser","locale",{"_92":93,"_94":95,"_20":128,"_139":49,"_140":49,"_141":142,"_146":147,"_169":-5,"_23":170,"_171":172,"_173":174,"_40":50,"_185":160,"_186":187,"_191":-5,"_192":-5,"_193":194,"_201":202,"_207":-5,"_181":156,"_208":209,"_213":214,"_8":404,"_405":406},"addedAt","2025-03-21T23:00:33.169169927Z","argumentsJsonSchema",{"_96":97,"_125":126,"_12":127},"properties",{"_98":99,"_102":103,"_105":106,"_108":109,"_113":114,"_116":117,"_119":120,"_122":123},"CAPS",{"_38":100,"_12":101},"Comma-separated list of capabilities to enable, possible values: tabs, pdf, history, wait, files, install","string","PORT",{"_38":104,"_12":101},"Port to listen on for SSE transport","VISION",{"_38":107,"_12":101},"Run server that uses screenshots (Aria snapshots are used by default)","BROWSER",{"_110":111,"_38":112,"_12":101},"default","chrome","Browser or chrome channel to use. Possible values: chrome, firefox, webkit, msedge, chrome-beta, chrome-canary, chrome-dev, msedge-beta, msedge-canary, msedge-dev","HEADLESS",{"_38":115,"_12":101},"Run browser in headless mode (headed by default)","CDP_ENDPOINT",{"_38":118,"_12":101},"CDP endpoint to connect to","USER_DATA_DIR",{"_38":121,"_12":101},"Path to the user data directory","EXECUTABLE_PATH",{"_38":124,"_12":101},"Path to the browser executable","required",[],"object",[129,130,131,132,133,134,135,136,137,138],"author:official","category:browser-automation","category:web-scraping","category:agent-orchestration","environment:local","hosting:local-only","language:typescript","security_grade:a","quality_grade:a","license_grade:a","descriptionMarkdown","descriptionPlainText","githubProject",{"_42":143,"_144":145},"## Playwright MCP\n\nA Model Context Protocol (MCP) server that provides browser automation capabilities using [Playwright](https://playwright.dev). This server enables LLMs to interact with web pages through structured accessibility snapshots, bypassing the need for screenshots or visually-tuned models.\n\n### Key Features\n\n- **Fast and lightweight**. Uses Playwright's accessibility tree, not pixel-based input.\n- **LLM-friendly**. No vision models needed, operates purely on structured data.\n- **Deterministic tool application**. Avoids ambiguity common with screenshot-based approaches.\n\n### Requirements\n- Node.js 18 or newer\n- VS Code, Cursor, Windsurf, Claude Desktop or any other MCP client\n\n\n\n### Getting started\n\nFirst, install the Playwright MCP server with your client. A typical configuration looks like this:\n\n```js\n{\n \"mcpServers\": {\n \"playwright\": {\n \"command\": \"npx\",\n \"args\": [\n \"@playwright/mcp@latest\"\n ]\n }\n }\n}\n```\n\n[ $\"Install$ ](https://insiders.vscode.dev/redirect?url=vscode%3Amcp%2Finstall%3F%257B%2522name%2522%253A%2522playwright%2522%252C%2522command%2522%253A%2522npx%2522%252C%2522args%2522%253A%255B%2522%2540playwright%252Fmcp%2540latest%2522%255D%257D) [ $\"Install$ ](https://insiders.vscode.dev/redirect?url=vscode-insiders%3Amcp%2Finstall%3F%257B%2522name%2522%253A%2522playwright%2522%252C%2522command%2522%253A%2522npx%2522%252C%2522args%2522%253A%255B%2522%2540playwright%252Fmcp%2540latest%2522%255D%257D)\n\n\n

Install in VS Code

\n\n

Install in Cursor

\n\n

Install in Windsurf

\n\n

Install in Claude Desktop

\n\n### Configuration\n\nPlaywright MCP server supports following arguments. They can be provided in the JSON configuration above, as a part of the `\"args\"` list:\n\n\n\n```\n> npx @playwright/mcp@latest --help\n --allowed-origins semicolon-separated list of origins to allow the\n browser to request. Default is to allow all.\n --blocked-origins semicolon-separated list of origins to block the\n browser from requesting. Blocklist is evaluated\n before allowlist. If used without the allowlist,\n requests not matching the blocklist are still\n allowed.\n --block-service-workers block service workers\n --browser browser or chrome channel to use, possible\n values: chrome, firefox, webkit, msedge.\n --caps comma-separated list of capabilities to enable,\n possible values: tabs, pdf, history, wait, files,\n install. Default is all.\n --cdp-endpoint CDP endpoint to connect to.\n --config path to the configuration file.\n --device device to emulate, for example: \"iPhone 15\"\n --executable-path path to the browser executable.\n --headless run browser in headless mode, headed by default\n --host host to bind server to. Default is localhost. Use\n 0.0.0.0 to bind to all interfaces.\n --ignore-https-errors ignore https errors\n --isolated keep the browser profile in memory, do not save\n it to disk.\n --image-responses whether to send image responses to the client.\n Can be \"allow\", \"omit\", or \"auto\". Defaults to\n \"auto\", which sends images if the client can\n display them.\n --no-sandbox disable the sandbox for all process types that\n are normally sandboxed.\n --output-dir path to the directory for output files.\n --port port to listen on for SSE transport.\n --proxy-bypass comma-separated domains to bypass proxy, for\n example \".com,chromium.org,.domain.com\"\n --proxy-server specify proxy server, for example\n \"http://myproxy:3128\" or \"socks5://myproxy:8080\"\n --save-trace Whether to save the Playwright Trace of the\n session into the output directory.\n --storage-state path to the storage state file for isolated\n sessions.\n --user-agent specify user agent string\n --user-data-dir path to the user data directory. If not\n specified, a temporary directory will be created.\n --viewport-size specify browser viewport size in pixels, for\n example \"1280, 720\"\n --vision Run server that uses screenshots (Aria snapshots\n are used by default)\n```\n\n\n\n### User profile\n\nYou can run Playwright MCP with persistent profile like a regular browser (default), or in the isolated contexts for the testing sessions.\n\n**Persistent profile**\n\nAll the logged in information will be stored in the persistent profile, you can delete it between sessions if you'd like to clear the offline state.\nPersistent profile is located at the following locations and you can override it with the `--user-data-dir` argument.\n\n```bash\n# Windows\n%USERPROFILE%\\AppData\\Local\\ms-playwright\\mcp-{channel}-profile\n\n# macOS\n- ~/Library/Caches/ms-playwright/mcp-{channel}-profile\n\n# Linux\n- ~/.cache/ms-playwright/mcp-{channel}-profile\n```\n\n**Isolated**\n\nIn the isolated mode, each session is started in the isolated profile. Every time you ask MCP to close the browser,\nthe session is closed and all the storage state for this session is lost. You can provide initial storage state\nto the browser via the config's `contextOptions` or via the `--storage-state` argument. Learn more about the storage\nstate [here](https://playwright.dev/docs/auth).\n\n```js\n{\n \"mcpServers\": {\n \"playwright\": {\n \"command\": \"npx\",\n \"args\": [\n \"@playwright/mcp@latest\",\n \"--isolated\",\n \"--storage-state={path/to/storage.json}\"\n ]\n }\n }\n}\n```\n\n### Configuration file\n\nThe Playwright MCP server can be configured using a JSON configuration file. You can specify the configuration file\nusing the `--config` command line option:\n\n```bash\nnpx @playwright/mcp@latest --config path/to/config.json\n```\n\n

Configuration file schema

Docker

\n\n

Programmatic usage

Interactions

\n\n

Navigation

\n\n

Resources

\n\n

Utilities

\n\n

Tabs

\n\n

Testing

\n\n

Vision mode

\n\n\n\n","url","https://github.com/microsoft/playwright-mcp","githubRepository",{"_148":149,"_150":151,"_152":153,"_154":155,"_40":156,"_157":158,"_162":163,"_164":165,"_167":168},"defaultBranch","main","fullName","microsoft/playwright-mcp","githubTreeId",122559,"language","TypeScript","playwright-mcp","owner",{"_159":160,"_8":161},"login","microsoft","wpw6ckTjuf","recentStargazers",433,"spdxLicense",{"_40":166},"Apache 2.0","stargazers",10793,"glamaJson",2886,"inspectable",true,"integrations",[175],{"_176":177,"_38":183,"_8":184},"brand",{"_178":179,"_40":180,"_181":182},"logoSvgPath","M12.504 0c-.155 0-.315.008-.48.021-4.226.333-3.105 4.807-3.17 6.298-.076 1.092-.3 1.953-1.05 3.02-.885 1.051-2.127 2.75-2.716 4.521-.278.832-.41 1.684-.287 2.489a.424.424 0 00-.11.135c-.26.268-.45.6-.663.839-.199.199-.485.267-.797.4-.313.136-.658.269-.864.68-.09.189-.136.394-.132.602 0 .199.027.4.055.536.058.399.116.728.04.97-.249.68-.28 1.145-.106 1.484.174.334.535.47.94.601.81.2 1.91.135 2.774.6.926.466 1.866.67 2.616.47.526-.116.97-.464 1.208-.946.587-.003 1.23-.269 2.26-.334.699-.058 1.574.267 2.577.2.025.134.063.198.114.333l.003.003c.391.778 1.113 1.132 1.884 1.071.771-.06 1.592-.536 2.257-1.306.631-.765 1.683-1.084 2.378-1.503.348-.199.629-.469.649-.853.023-.4-.2-.811-.714-1.376v-.097l-.003-.003c-.17-.2-.25-.535-.338-.926-.085-.401-.182-.786-.492-1.046h-.003c-.059-.054-.123-.067-.188-.135a.357.357 0 00-.19-.064c.431-1.278.264-2.55-.173-3.694-.533-1.41-1.465-2.638-2.175-3.483-.796-1.005-1.576-1.957-1.56-3.368.026-2.152.236-6.133-3.544-6.139zm.529 3.405h.013c.213 0 .396.062.584.198.19.135.33.332.438.533.105.259.158.459.166.724 0-.02.006-.04.006-.06v.105a.086.086 0 01-.004-.021l-.004-.024a1.807 1.807 0 01-.15.706.953.953 0 01-.213.335.71.71 0 00-.088-.042c-.104-.045-.198-.064-.284-.133a1.312 1.312 0 00-.22-.066c.05-.06.146-.133.183-.198.053-.128.082-.264.088-.402v-.02a1.21 1.21 0 00-.061-.4c-.045-.134-.101-.2-.183-.333-.084-.066-.167-.132-.267-.132h-.016c-.093 0-.176.03-.262.132a.8.8 0 00-.205.334 1.18 1.18 0 00-.09.4v.019c.002.089.008.179.02.267-.193-.067-.438-.135-.607-.202a1.635 1.635 0 01-.018-.2v-.02a1.772 1.772 0 01.15-.768c.082-.22.232-.406.43-.533a.985.985 0 01.594-.2zm-2.962.059h.036c.142 0 .27.048.399.135.146.129.264.288.344.465.09.199.14.4.153.667v.004c.007.134.006.2-.002.266v.08c-.03.007-.056.018-.083.024-.152.055-.274.135-.393.2.012-.09.013-.18.003-.267v-.015c-.012-.133-.04-.2-.082-.333a.613.613 0 00-.166-.267.248.248 0 00-.183-.064h-.021c-.071.006-.13.04-.186.132a.552.552 0 00-.12.27.944.944 0 00-.023.33v.015c.012.135.037.2.08.334.046.134.098.2.166.268.01.009.02.018.034.024-.07.057-.117.07-.176.136a.304.304 0 01-.131.068 2.62 2.62 0 01-.275-.402 1.772 1.772 0 01-.155-.667 1.759 1.759 0 01.08-.668 1.43 1.43 0 01.283-.535c.128-.133.26-.2.418-.2zm1.37 1.706c.332 0 .733.065 1.216.399.293.2.523.269 1.052.468h.003c.255.136.405.266.478.399v-.131a.571.571 0 01.016.47c-.123.31-.516.643-1.063.842v.002c-.268.135-.501.333-.775.465-.276.135-.588.292-1.012.267a1.139 1.139 0 01-.448-.067 3.566 3.566 0 01-.322-.198c-.195-.135-.363-.332-.612-.465v-.005h-.005c-.4-.246-.616-.512-.686-.71-.07-.268-.005-.47.193-.6.224-.135.38-.271.483-.336.104-.074.143-.102.176-.131h.002v-.003c.169-.202.436-.47.839-.601.139-.036.294-.065.466-.065zm2.8 2.142c.358 1.417 1.196 3.475 1.735 4.473.286.534.855 1.659 1.102 3.024.156-.005.33.018.513.064.646-1.671-.546-3.467-1.089-3.966-.22-.2-.232-.335-.123-.335.59.534 1.365 1.572 1.646 2.757.13.535.16 1.104.021 1.67.067.028.135.06.205.067 1.032.534 1.413.938 1.23 1.537v-.043c-.06-.003-.12 0-.18 0h-.016c.151-.467-.182-.825-1.065-1.224-.915-.4-1.646-.336-1.77.465-.008.043-.013.066-.018.135-.068.023-.139.053-.209.064-.43.268-.662.669-.793 1.187-.13.533-.17 1.156-.205 1.869v.003c-.02.334-.17.838-.319 1.35-1.5 1.072-3.58 1.538-5.348.334a2.645 2.645 0 00-.402-.533 1.45 1.45 0 00-.275-.333c.182 0 .338-.03.465-.067a.615.615 0 00.314-.334c.108-.267 0-.697-.345-1.163-.345-.467-.931-.995-1.788-1.521-.63-.4-.986-.87-1.15-1.396-.165-.534-.143-1.085-.015-1.645.245-1.07.873-2.11 1.274-2.763.107-.065.037.135-.408.974-.396.751-1.14 2.497-.122 3.854a8.123 8.123 0 01.647-2.876c.564-1.278 1.743-3.504 1.836-5.268.048.036.217.135.289.202.218.133.38.333.59.465.21.201.477.335.876.335.039.003.075.006.11.006.412 0 .73-.134.997-.268.29-.134.52-.334.74-.4h.005c.467-.135.835-.402 1.044-.7zm2.185 8.958c.037.6.343 1.245.882 1.377.588.134 1.434-.333 1.791-.765l.211-.01c.315-.007.577.01.847.268l.003.003c.208.199.305.53.391.876.085.4.154.78.409 1.066.486.527.645.906.636 1.14l.003-.007v.018l-.003-.012c-.015.262-.185.396-.498.595-.63.401-1.746.712-2.457 1.57-.618.737-1.37 1.14-2.036 1.191-.664.053-1.237-.2-1.574-.898l-.005-.003c-.21-.4-.12-1.025.056-1.69.176-.668.428-1.344.463-1.897.037-.714.076-1.335.195-1.814.12-.465.308-.797.641-.984l.045-.022zm-10.814.049h.01c.053 0 .105.005.157.014.376.055.706.333 1.023.752l.91 1.664.003.003c.243.533.754 1.064 1.189 1.637.434.598.77 1.131.729 1.57v.006c-.057.744-.48 1.148-1.125 1.294-.645.135-1.52.002-2.395-.464-.968-.536-2.118-.469-2.857-.602-.369-.066-.61-.2-.723-.4-.11-.2-.113-.602.123-1.23v-.004l.002-.003c.117-.334.03-.752-.027-1.118-.055-.401-.083-.71.043-.94.16-.334.396-.4.69-.533.294-.135.64-.202.915-.47h.002v-.002c.256-.268.445-.601.668-.838.19-.201.38-.336.663-.336zm7.159-9.074c-.435.201-.945.535-1.488.535-.542 0-.97-.267-1.28-.466-.154-.134-.28-.268-.373-.335-.164-.134-.144-.333-.074-.333.109.016.129.134.199.2.096.066.215.2.36.333.292.2.68.467 1.167.467.485 0 1.053-.267 1.398-.466.195-.135.445-.334.648-.467.156-.136.149-.267.279-.267.128.016.034.134-.147.332a8.097 8.097 0 01-.69.468zm-1.082-1.583V5.64c-.006-.02.013-.042.029-.05.074-.043.18-.027.26.004.063 0 .16.067.15.135-.006.049-.085.066-.135.066-.055 0-.092-.043-.141-.068-.052-.018-.146-.008-.163-.065zm-.551 0c-.02.058-.113.049-.166.066-.047.025-.086.068-.14.068-.05 0-.13-.02-.136-.068-.01-.066.088-.133.15-.133.08-.031.184-.047.259-.005.019.009.036.03.03.05v.02h.003z","Linux","slug","linux","Provides browser automation capabilities on Linux systems without a display by running Playwright in a client-server manner, allowing for headed browser operations in display-less environments.","xkihip2thw","namespace","npmPackage",{"_40":188,"_189":190},"@playwright/mcp","weeklyDownloads",47061,"prompts","redditPostForeignId","resources",[195],{"_196":197,"_40":198,"_199":200},"uri","browser://console","Page console","mimeType","text/plain","scores",{"_203":204,"_205":204,"_206":204},"license",100,"quality","security","simpleIcon","supportedPlatforms",[210,211,212],"WINDOWS","LINUX","MACOS","tools",[215,225,235,247,259,264,273,281,286,291,296,301,313,330,338,356,368,376,381,388,397],{"_40":216,"_38":217,"_218":219},"browser_close","Close the page","inputSchema",{"_12":127,"_220":221,"_96":222,"_223":224},"$schema","http://json-schema.org/draft-07/schema#",{},"additionalProperties",false,{"_40":226,"_38":227,"_218":228},"browser_wait","Wait for a specified time in seconds",{"_12":127,"_220":221,"_125":229,"_96":231,"_223":224},[230],"time",{"_230":232},{"_12":233,"_38":234},"number","The time to wait in seconds",{"_40":236,"_38":237,"_218":238},"browser_resize","Resize the browser window",{"_12":127,"_220":221,"_125":239,"_96":242,"_223":224},[240,241],"width","height",{"_240":243,"_241":245},{"_12":233,"_38":244},"Width of the browser window",{"_12":233,"_38":246},"Height of the browser window",{"_40":248,"_38":249,"_218":250},"browser_file_upload","Upload one or multiple files",{"_12":127,"_220":221,"_125":251,"_96":253,"_223":224},[252],"paths",{"_252":254},{"_12":255,"_256":257,"_38":258},"array","items",{"_12":101},"The absolute paths to the files to upload. Can be a single file or multiple files.",{"_40":260,"_38":261,"_218":262},"browser_install","Install the browser specified in the config. Call this if you get an error about the browser not being installed.",{"_12":127,"_220":221,"_96":263,"_223":224},{},{"_40":265,"_38":266,"_218":267},"browser_press_key","Press a key on the keyboard",{"_12":127,"_220":221,"_125":268,"_96":270,"_223":224},[269],"key",{"_269":271},{"_12":101,"_38":272},"Name of the key to press or a character to generate, such as `ArrowLeft` or `a`",{"_40":274,"_38":275,"_218":276},"browser_navigate","Navigate to a URL",{"_12":127,"_220":221,"_125":277,"_96":278,"_223":224},[144],{"_144":279},{"_12":101,"_38":280},"The URL to navigate to",{"_40":282,"_38":283,"_218":284},"browser_navigate_back","Go back to the previous page",{"_12":127,"_220":221,"_96":285,"_223":224},{},{"_40":287,"_38":288,"_218":289},"browser_navigate_forward","Go forward to the next page",{"_12":127,"_220":221,"_96":290,"_223":224},{},{"_40":292,"_38":293,"_218":294},"browser_pdf_save","Save page as PDF",{"_12":127,"_220":221,"_96":295,"_223":224},{},{"_40":297,"_38":298,"_218":299},"browser_snapshot","Capture accessibility snapshot of the current page, this is better than screenshot",{"_12":127,"_220":221,"_96":300,"_223":224},{},{"_40":302,"_38":303,"_218":304},"browser_click","Perform click on a web page",{"_12":127,"_220":221,"_125":305,"_96":308,"_223":224},[306,307],"element","ref",{"_307":309,"_306":311},{"_12":101,"_38":310},"Exact target element reference from the page snapshot",{"_12":101,"_38":312},"Human-readable element description used to obtain permission to interact with the element",{"_40":314,"_38":315,"_218":316},"browser_drag","Perform drag and drop between two elements",{"_12":127,"_220":221,"_125":317,"_96":322,"_223":224},[318,319,320,321],"startElement","startRef","endElement","endRef",{"_321":323,"_319":324,"_320":326,"_318":328},{"_12":101,"_38":310},{"_12":101,"_38":325},"Exact source element reference from the page snapshot",{"_12":101,"_38":327},"Human-readable target element description used to obtain the permission to interact with the element",{"_12":101,"_38":329},"Human-readable source element description used to obtain the permission to interact with the element",{"_40":331,"_38":332,"_218":333},"browser_hover","Hover over element on page",{"_12":127,"_220":221,"_125":334,"_96":335,"_223":224},[306,307],{"_307":336,"_306":337},{"_12":101,"_38":310},{"_12":101,"_38":312},{"_40":339,"_38":340,"_218":341},"browser_type","Type text into editable element",{"_12":127,"_220":221,"_125":342,"_96":344,"_223":224},[306,307,343],"text",{"_307":345,"_343":346,"_348":349,"_352":353,"_306":355},{"_12":101,"_38":310},{"_12":101,"_38":347},"Text to type into the element","slowly",{"_12":350,"_38":351},"boolean","Whether to type one character at a time. Useful for triggering key handlers in the page. By default entire text is filled in at once.","submit",{"_12":350,"_38":354},"Whether to submit entered text (press Enter after)",{"_12":101,"_38":312},{"_40":357,"_38":358,"_218":359},"browser_select_option","Select an option in a dropdown",{"_12":127,"_220":221,"_125":360,"_96":362,"_223":224},[306,307,361],"values",{"_307":363,"_361":364,"_306":367},{"_12":101,"_38":310},{"_12":255,"_256":365,"_38":366},{"_12":101},"Array of values to select in the dropdown. This can be a single value or multiple values.",{"_12":101,"_38":312},{"_40":369,"_38":370,"_218":371},"browser_take_screenshot","Take a screenshot of the current page. You can't perform actions based on the screenshot, use browser_snapshot for actions.",{"_12":127,"_220":221,"_96":372,"_223":224},{"_373":374},"raw",{"_12":350,"_38":375},"Whether to return without compression (in PNG format). Default is false, which returns a JPEG image.",{"_40":377,"_38":378,"_218":379},"browser_tab_list","List browser tabs",{"_12":127,"_220":221,"_96":380,"_223":224},{},{"_40":382,"_38":383,"_218":384},"browser_tab_new","Open a new tab",{"_12":127,"_220":221,"_96":385,"_223":224},{"_144":386},{"_12":101,"_38":387},"The URL to navigate to in the new tab. If not provided, the new tab will be blank.",{"_40":389,"_38":390,"_218":391},"browser_tab_select","Select a tab by index",{"_12":127,"_220":221,"_125":392,"_96":394,"_223":224},[393],"index",{"_393":395},{"_12":233,"_38":396},"The index of the tab to select",{"_40":398,"_38":399,"_218":400},"browser_tab_close","Close a tab",{"_12":127,"_220":221,"_96":401,"_223":224},{"_393":402},{"_12":233,"_38":403},"The index of the tab to close. Closes current tab if not provided.","92mrqijm10","updatedAt","2025-05-29T01:00:33.862862894Z","meta",[409,411,413,420,423,425,428,431,434,437,440,442,445,448,450,452,454,456,458],{"_10":410},"Playwright MCP Server | Glama",{"_412":49,"_40":38},"content",{"_414":415,"_416":417,"_418":419},"href","https://glama.ai/mcp/servers/@microsoft/playwright-mcp","rel","canonical","tagName","link",{"_412":50,"_421":422},"property","og:title",{"_412":49,"_421":424},"og:description",{"_412":426,"_421":427},"https://glama.ai/logo.png","og:logo",{"_412":429,"_421":430},"https://glama.ai/mcp/servers/@microsoft/playwright-mcp/og-image","og:image",{"_412":432,"_421":433},"630","og:image:height",{"_412":435,"_421":436},"1200","og:image:width",{"_412":438,"_421":439},"website","og:type",{"_412":415,"_421":441},"og:url",{"_412":443,"_421":444},"Glama – MCP Hosting Platform","og:site_name",{"_412":446,"_40":447},"summary_large_image","twitter:card",{"_412":443,"_40":449},"twitter:title",{"_412":49,"_40":451},"twitter:description",{"_412":429,"_40":453},"twitter:image",{"_412":432,"_40":455},"twitter:image:height",{"_412":435,"_40":457},"twitter:image:width",{"_459":460},"script:ld+json",{"_461":462,"_463":464},"@context","http://schema.org","@graph",[465],{"_461":466,"_467":468,"_469":470},"https://schema.org","@type","BreadcrumbList","itemListElement",[471,478,482,486],{"_467":472,"_473":474,"_40":475,"_476":477},"ListItem","item","https://glama.ai/","Glama","position",1,{"_467":472,"_473":479,"_40":480,"_476":481},"https://glama.ai/mcp","MCP",2,{"_467":472,"_473":483,"_40":484,"_476":485},"https://glama.ai/mcp/servers","Servers",3,{"_467":472,"_473":415,"_40":50,"_476":487},4,"visitorIsMaintainer","routes/_public/mcp/servers/~namespace/~slug/_pages/reviews/_route",{"_89":-5,"_36":491,"_407":619,"_654":-5},{"_92":93,"_94":492,"_20":503,"_139":49,"_140":49,"_141":504,"_146":505,"_169":-5,"_23":170,"_171":172,"_173":508,"_40":50,"_185":160,"_186":511,"_191":-5,"_192":-5,"_193":512,"_201":514,"_207":-5,"_181":156,"_208":515,"_213":516,"_8":404,"_405":406},{"_96":493,"_125":502,"_12":127},{"_98":494,"_102":495,"_105":496,"_108":497,"_113":498,"_116":499,"_119":500,"_122":501},{"_38":100,"_12":101},{"_38":104,"_12":101},{"_38":107,"_12":101},{"_110":111,"_38":112,"_12":101},{"_38":115,"_12":101},{"_38":118,"_12":101},{"_38":121,"_12":101},{"_38":124,"_12":101},[],[129,130,131,132,133,134,135,136,137,138],{"_42":143,"_144":145},{"_148":149,"_150":151,"_152":153,"_154":155,"_40":156,"_157":506,"_162":163,"_164":507,"_167":168},{"_159":160,"_8":161},{"_40":166},[509],{"_176":510,"_38":183,"_8":184},{"_178":179,"_40":180,"_181":182},{"_40":188,"_189":190},[513],{"_196":197,"_40":198,"_199":200},{"_203":204,"_205":204,"_206":204},[210,211,212],[517,520,525,531,537,540,545,550,553,556,559,562,568,576,582,591,599,603,606,610,615],{"_40":216,"_38":217,"_218":518},{"_12":127,"_220":221,"_96":519,"_223":224},{},{"_40":226,"_38":227,"_218":521},{"_12":127,"_220":221,"_125":522,"_96":523,"_223":224},[230],{"_230":524},{"_12":233,"_38":234},{"_40":236,"_38":237,"_218":526},{"_12":127,"_220":221,"_125":527,"_96":528,"_223":224},[240,241],{"_240":529,"_241":530},{"_12":233,"_38":244},{"_12":233,"_38":246},{"_40":248,"_38":249,"_218":532},{"_12":127,"_220":221,"_125":533,"_96":534,"_223":224},[252],{"_252":535},{"_12":255,"_256":536,"_38":258},{"_12":101},{"_40":260,"_38":261,"_218":538},{"_12":127,"_220":221,"_96":539,"_223":224},{},{"_40":265,"_38":266,"_218":541},{"_12":127,"_220":221,"_125":542,"_96":543,"_223":224},[269],{"_269":544},{"_12":101,"_38":272},{"_40":274,"_38":275,"_218":546},{"_12":127,"_220":221,"_125":547,"_96":548,"_223":224},[144],{"_144":549},{"_12":101,"_38":280},{"_40":282,"_38":283,"_218":551},{"_12":127,"_220":221,"_96":552,"_223":224},{},{"_40":287,"_38":288,"_218":554},{"_12":127,"_220":221,"_96":555,"_223":224},{},{"_40":292,"_38":293,"_218":557},{"_12":127,"_220":221,"_96":558,"_223":224},{},{"_40":297,"_38":298,"_218":560},{"_12":127,"_220":221,"_96":561,"_223":224},{},{"_40":302,"_38":303,"_218":563},{"_12":127,"_220":221,"_125":564,"_96":565,"_223":224},[306,307],{"_307":566,"_306":567},{"_12":101,"_38":310},{"_12":101,"_38":312},{"_40":314,"_38":315,"_218":569},{"_12":127,"_220":221,"_125":570,"_96":571,"_223":224},[318,319,320,321],{"_321":572,"_319":573,"_320":574,"_318":575},{"_12":101,"_38":310},{"_12":101,"_38":325},{"_12":101,"_38":327},{"_12":101,"_38":329},{"_40":331,"_38":332,"_218":577},{"_12":127,"_220":221,"_125":578,"_96":579,"_223":224},[306,307],{"_307":580,"_306":581},{"_12":101,"_38":310},{"_12":101,"_38":312},{"_40":339,"_38":340,"_218":583},{"_12":127,"_220":221,"_125":584,"_96":585,"_223":224},[306,307,343],{"_307":586,"_343":587,"_348":588,"_352":589,"_306":590},{"_12":101,"_38":310},{"_12":101,"_38":347},{"_12":350,"_38":351},{"_12":350,"_38":354},{"_12":101,"_38":312},{"_40":357,"_38":358,"_218":592},{"_12":127,"_220":221,"_125":593,"_96":594,"_223":224},[306,307,361],{"_307":595,"_361":596,"_306":598},{"_12":101,"_38":310},{"_12":255,"_256":597,"_38":366},{"_12":101},{"_12":101,"_38":312},{"_40":369,"_38":370,"_218":600},{"_12":127,"_220":221,"_96":601,"_223":224},{"_373":602},{"_12":350,"_38":375},{"_40":377,"_38":378,"_218":604},{"_12":127,"_220":221,"_96":605,"_223":224},{},{"_40":382,"_38":383,"_218":607},{"_12":127,"_220":221,"_96":608,"_223":224},{"_144":609},{"_12":101,"_38":387},{"_40":389,"_38":390,"_218":611},{"_12":127,"_220":221,"_125":612,"_96":613,"_223":224},[393],{"_393":614},{"_12":233,"_38":396},{"_40":398,"_38":399,"_218":616},{"_12":127,"_220":221,"_96":617,"_223":224},{"_393":618},{"_12":233,"_38":403},[620,622,624,626,628,629,630,631,632,633,634,635,636,637,638,639,640,641,642],{"_10":621},"Reviews | Playwright MCP Server | Glama",{"_412":623,"_40":38},"Read community reviews and feedback for Playwright MCP Server. Discover user experiences, ratings, strengths, and potential limitations of this MCP implementation from actual users.",{"_414":625,"_416":417,"_418":419},"https://glama.ai/mcp/servers/@microsoft/playwright-mcp/reviews",{"_412":627,"_421":422},"Reviews | Playwright MCP Server",{"_412":623,"_421":424},{"_412":426,"_421":427},{"_412":429,"_421":430},{"_412":432,"_421":433},{"_412":435,"_421":436},{"_412":438,"_421":439},{"_412":625,"_421":441},{"_412":443,"_421":444},{"_412":446,"_40":447},{"_412":443,"_40":449},{"_412":623,"_40":451},{"_412":429,"_40":453},{"_412":432,"_40":455},{"_412":435,"_40":457},{"_459":643},{"_461":462,"_463":644},[645],{"_461":466,"_467":468,"_469":646},[647,648,649,650,651],{"_467":472,"_473":474,"_40":475,"_476":477},{"_467":472,"_473":479,"_40":480,"_476":481},{"_467":472,"_473":483,"_40":484,"_476":485},{"_467":472,"_473":415,"_40":50,"_476":487},{"_467":472,"_473":625,"_40":652,"_476":653},"Reviews",5,"review","actionData","errors"]

Rate & Review

MCP directory API