The Playwright MCP Server provides comprehensive browser automation capabilities, enabling interaction with web pages using structured accessibility snapshots rather than pixel-based inputs. This server allows you to:
Capture accessibility snapshots of web pages
Interact with page elements (click, hover, type, drag and drop, select options)
Navigate between pages and manage browser tabs
Wait for specific conditions or timeouts
Handle file uploads and browser dialogs
Capture screenshots and save PDFs
Monitor network requests and console messages
Generate Playwright test scripts
It also offers a Vision Mode for coordinate-based interactions like screen capture, mouse movement, clicking, and typing.
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.
Click on "Install Server".
Wait a few minutes for the server to deploy. Once ready, it will show a "Started" state.
In the chat, type @ followed by the MCP server name and your instructions, e.g., "@Playwright MCP Server take a snapshot of the GitHub homepage and show me the main navigation links"
That's it! The server will respond to your query, and you can continue using it as needed.
Here is a step-by-step guide with screenshots.
A Model Context Protocol (MCP) server that provides browser automation capabilities using Playwright. This server enables LLMs to interact with web pages through structured accessibility snapshots, bypassing the need for screenshots or visually-tuned models.
This package provides MCP interface into Playwright. If you are using a coding agent, you might benefit from using the CLI+SKILLS instead.
CLI: Modern coding agents increasingly favor CLI–based workflows exposed as SKILLs over MCP because CLI invocations are more token-efficient: they avoid loading large tool schemas and verbose accessibility trees into the model context, allowing agents to act through concise, purpose-built commands. This makes CLI + SKILLs better suited for high-throughput coding agents that must balance browser automation with large codebases, tests, and reasoning within limited context windows.Learn more about Playwright CLI with SKILLS.
MCP: MCP remains relevant for specialized agentic loops that benefit from persistent state, rich introspection, and iterative reasoning over page structure, such as exploratory automation, self-healing tests, or long-running autonomous workflows where maintaining continuous browser context outweighs token cost concerns.
Fast and lightweight. Uses Playwright's accessibility tree, not pixel-based input.
LLM-friendly. No vision models needed, operates purely on structured data.
Deterministic tool application. Avoids ambiguity common with screenshot-based approaches.
Node.js 18 or newer
VS Code, Cursor, Windsurf, Claude Desktop, Goose, Grok, Junie or any other MCP client
First, install the Playwright MCP server with your client.
Standard config works in most of the tools:
{
"mcpServers": {
"playwright": {
"command": "npx",
"args": [
"@playwright/mcp@latest"
]
}
}
}
Add via the Amp VS Code extension settings screen or by updating your settings.json file:
"amp.mcpServers": {
"playwright": {
"command": "npx",
"args": [
"@playwright/mcp@latest"
]
}
}Amp CLI Setup:
Add via the amp mcp add command below
amp mcp add playwright -- npx @playwright/mcp@latestAdd via the Antigravity settings or by updating your configuration file:
{
"mcpServers": {
"playwright": {
"command": "npx",
"args": [
"@playwright/mcp@latest"
]
}
}
}Use the Claude Code CLI to add the Playwright MCP server:
claude mcp add playwright npx @playwright/mcp@latestFollow the MCP install guide, use the standard config above.
Follow the instruction in the section Configuring MCP Servers
Example: Local Setup
Add the following to your cline_mcp_settings.json file:
{
"mcpServers": {
"playwright": {
"type": "stdio",
"command": "npx",
"timeout": 30,
"args": [
"-y",
"@playwright/mcp@latest"
],
"disabled": false
}
}
}Use the Codex CLI to add the Playwright MCP server:
codex mcp add playwright npx "@playwright/mcp@latest"Alternatively, create or edit the configuration file ~/.codex/config.toml and add:
[mcp_servers.playwright]
command = "npx"
args = ["@playwright/mcp@latest"]For more information, see the Codex MCP documentation.
Use the Copilot CLI to interactively add the Playwright MCP server:
/mcp addAlternatively, create or edit the configuration file ~/.copilot/mcp-config.json and add:
{
"mcpServers": {
"playwright": {
"type": "local",
"command": "npx",
"tools": [
"*"
],
"args": [
"@playwright/mcp@latest"
]
}
}
}For more information, see the Copilot CLI documentation.
Go to Cursor Settings -> MCP -> Add new MCP Server. Name to your liking, use command type with the command npx @playwright/mcp@latest. You can also verify config or add command like arguments via clicking Edit.
Use the Factory CLI to add the Playwright MCP server:
droid mcp add playwright "npx @playwright/mcp@latest"Alternatively, type /mcp within Factory droid to open an interactive UI for managing MCP servers.
For more information, see the Factory MCP documentation.
Follow the MCP install guide, use the standard config above.
Go to Advanced settings -> Extensions -> Add custom extension. Name to your liking, use type STDIO, and set the command to npx @playwright/mcp. Click "Add Extension".
Use the Grok CLI to add the Playwright MCP server:
grok mcp add playwright -- npx @playwright/mcp@latestAlternatively, create or edit the configuration file ~/.grok/config.toml and add:
[mcp_servers.playwright]
command = "npx"
args = ["@playwright/mcp@latest"]For more information, see the Grok MCP documentation.
To add the Playwright MCP server in Junie CLI:
Type /mcp
Press Ctrl+A to add a new MCP server
Select Playwright from the list
Alternatively, add to .junie/mcp/mcp.json:
{
"mcpServers": {
"Playwright": {
"command": "npx",
"args": [
"-y",
"@playwright/mcp@latest"
]
}
}
}For more information, see the Junie MCP configuration documentation.
Follow the MCP Servers documentation. For example in .kiro/settings/mcp.json:
{
"mcpServers": {
"playwright": {
"command": "npx",
"args": [
"@playwright/mcp@latest"
]
}
}
}
Go to Program in the right sidebar -> Install -> Edit mcp.json. Use the standard config above.
Follow the MCP Servers documentation. For example in ~/.config/opencode/opencode.json:
{
"$schema": "https://opencode.ai/config.json",
"mcp": {
"playwright": {
"type": "local",
"command": [
"npx",
"@playwright/mcp@latest"
],
"enabled": true
}
}
}
Open Qodo Gen chat panel in VSCode or IntelliJ → Connect more tools → + Add new MCP → Paste the standard config above.
Click Save.
Follow the MCP install guide, use the standard config above. You can also install the Playwright MCP server using the VS Code CLI:
# For VS Code
code --add-mcp '{"name":"playwright","command":"npx","args":["@playwright/mcp@latest"]}'After installation, the Playwright MCP server will be available for use with your GitHub Copilot agent in VS Code.
Go to Settings -> AI -> Manage MCP Servers -> + Add to add an MCP Server. Use the standard config above.
Alternatively, use the slash command /add-mcp in the Warp prompt and paste the standard config from above:
{
"mcpServers": {
"playwright": {
"command": "npx",
"args": [
"@playwright/mcp@latest"
]
}
}
}Follow Windsurf MCP documentation. Use the standard config above.
Playwright MCP server supports following arguments. They can be provided in the JSON configuration above, as a part of the "args" list:
Option | Description |
--allowed-hosts <hosts...> | comma-separated list of hosts this server is allowed to serve from. Defaults to the host the server is bound to. Pass '*' to disable the host check.env |
--allowed-origins | semicolon-separated list of TRUSTED origins to allow the browser to request. Default is to allow all. Important: does not serve as a security boundary and does not affect redirects.env |
--allow-unrestricted-file-access | allow access to files outside of the workspace roots. Also allows unrestricted access to file:// URLs. By default access to file system is restricted to workspace root directories (or cwd if no roots are configured) only, and navigation to file:// URLs is blocked.env |
--blocked-origins | semicolon-separated list of origins to block the browser from requesting. Blocklist is evaluated before allowlist. If used without the allowlist, requests not matching the blocklist are still allowed. Important: does not serve as a security boundary and does not affect redirects.env |
--block-service-workers | block service workersenv |
--browser | browser or chrome channel to use, possible values: chrome, firefox, webkit, msedge.env |
--caps | comma-separated list of additional capabilities to enable, possible values: vision, pdf, devtools.env |
--cdp-endpoint | CDP endpoint to connect to.env |
--cdp-header <headers...> | CDP headers to send with the connect request, multiple can be specified.env |
--cdp-timeout | timeout in milliseconds for connecting to CDP endpoint, defaults to 30000msenv |
--codegen | specify the language to use for code generation, possible values: "typescript", "python", "java", "csharp", "none". Default is "typescript".env |
--config | path to the configuration file.env |
--console-level | level of console messages to return: "error", "warning", "info", "debug". Each level includes the messages of more severe levels.env |
--device | device to emulate, for example: "iPhone 15"env |
--mobile | emulate a generic mobile device (Pixel 10 for Chromium, iPhone 17 for WebKit). Mobile pages are usually lighter, which saves tokens. Cannot be combined with --device.env |
--executable-path | path to the browser executable.env |
--extension | Connect to a running browser instance (Edge/Chrome only). Requires the "Playwright Extension" to be installed.env |
--endpoint | Bound browser endpoint to connect to.env |
--grant-permissions <permissions...> | List of permissions to grant to the browser context, for example "geolocation", "clipboard-read", "clipboard-write".env |
--headless | run browser in headless mode, headed by defaultenv |
--host | host to bind server to. Default is localhost. Use 0.0.0.0 to bind to all interfaces.env |
--ignore-https-errors | ignore https errorsenv |
--init-page <path...> | path to TypeScript file to evaluate on Playwright page objectenv |
--init-script <path...> | path to JavaScript file to add as an initialization script. The script will be evaluated in every page before any of the page's scripts. Can be specified multiple times.env |
--isolated | keep the browser profile in memory, do not save it to disk.env |
--image-responses | whether to send image responses to the client. Can be "allow" or "omit", Defaults to "allow".env |
--no-sandbox | disable the sandbox for all process types that are normally sandboxed.env |
--output-dir | path to the directory for output files.env |
--output-max-size | Threshold for evicting old output files, in bytes.env |
--port | port to listen on for SSE transport.env |
--proxy-bypass | comma-separated domains to bypass proxy, for example ".com,chromium.org,.domain.com"env |
--proxy-server | specify proxy server, for example "http://myproxy:3128" or "socks5://myproxy:8080"env |
--sandbox | enable the sandbox for all process types that are normally not sandboxed.env |
--save-session | Whether to save the Playwright MCP session into the output directory.env |
--secrets | path to a file containing secrets in the dotenv formatenv |
--shared-browser-context | reuse the same browser context between all connected HTTP clients.env |
--snapshot-boxes | include each element's bounding box as [box=x,y,width,height] in snapshots. Coordinates are viewport-relative, in CSS pixels.env |
--snapshot-mode | when taking snapshots for responses, specifies the mode to use. Can be "full" or "none". Default is "full".env |
--storage-state | path to the storage state file for isolated sessions.env |
--test-id-attribute | specify the attribute to use for test ids, defaults to "data-testid"env |
--timeout-action | specify action timeout in milliseconds, defaults to 5000msenv |
--timeout-navigation | specify navigation timeout in milliseconds, defaults to 60000msenv |
--timeout-settle | how long to wait after each action for triggered work to settle, in milliseconds, defaults to 500msenv |
--user-agent | specify user agent stringenv |
--user-data-dir | path to the user data directory. If not specified, a temporary directory will be created.env |
--viewport-size | specify browser viewport size in pixels, for example "1280x720"env |
You can run Playwright MCP with persistent profile like a regular browser (default), in isolated contexts for testing sessions, or connect to your existing browser using the browser extension.
Persistent profile
All 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.
Persistent profile is located at the following locations and you can override it with the --user-data-dir argument.
# Windows
%USERPROFILE%\AppData\Local\ms-playwright\mcp-{channel}-{workspace-hash}
# macOS
- ~/Library/Caches/ms-playwright/mcp-{channel}-{workspace-hash}
# Linux
- ~/.cache/ms-playwright/mcp-{channel}-{workspace-hash}{workspace-hash} is derived from the MCP client's workspace root, so different projects get separate profiles automatically.
A persistent profile can only be used by one browser instance at a time, so concurrent MCP clients sharing the same workspace will conflict. To run several clients in parallel, start each additional client with--isolated or point it at a distinct --user-data-dir.
Isolated
In the isolated mode, each session is started in the isolated profile. Every time you ask MCP to close the browser,
the session is closed and all the storage state for this session is lost. You can provide initial storage state
to the browser via the config's contextOptions or via the --storage-state argument. Learn more about the storage
state here.
{
"mcpServers": {
"playwright": {
"command": "npx",
"args": [
"@playwright/mcp@latest",
"--isolated",
"--storage-state={path/to/storage.json}"
]
}
}
}Browser Extension
The Playwright MCP Chrome Extension allows you to connect to existing browser tabs and leverage your logged-in sessions and browser state. See microsoft/playwright › packages/extension for installation and setup instructions.
There are multiple ways to provide the initial state to the browser context or a page.
For the storage state, you can either:
Start with a user data directory using the --user-data-dir argument. This will persist all browser data between the sessions.
Start with a storage state file using the --storage-state argument. This will load cookies and local storage from the file into an isolated browser context.
For the page state, you can use:
--init-page to point to a TypeScript file that will be evaluated on the Playwright page object. This allows you to run arbitrary code to set up the page.
// init-page.ts
export default async ({ page }) => {
await page.context().grantPermissions(['geolocation']);
await page.context().setGeolocation({ latitude: 37.7749, longitude: -122.4194 });
await page.setViewportSize({ width: 1280, height: 720 });
};--init-script to point to a JavaScript file that will be added as an initialization script. The script will be evaluated in every page before any of the page's scripts.
This is useful for overriding browser APIs or setting up the environment.
// init-script.js
window.isPlaywrightMCP = true;The Playwright MCP server can be configured using a JSON configuration file. You can specify the configuration file
using the --config command line option:
npx @playwright/mcp@latest --config path/to/config.json{
/**
* The browser to use.
*/
browser?: {
/**
* The type of browser to use.
*/
browserName?: 'chromium' | 'firefox' | 'webkit';
/**
* Keep the browser profile in memory, do not save it to disk.
*/
isolated?: boolean;
/**
* Path to a user data directory for browser profile persistence.
* Temporary directory is created by default.
*/
userDataDir?: string;
/**
* Launch options passed to
* @see https://playwright.dev/docs/api/class-browsertype#browser-type-launch-persistent-context
*
* This is useful for settings options like `channel`, `headless`, `executablePath`, etc.
*/
launchOptions?: playwright.LaunchOptions;
/**
* Context options for the browser context.
*
* This is useful for settings options like `viewport`.
*/
contextOptions?: playwright.BrowserContextOptions;
/**
* Chrome DevTools Protocol endpoint to connect to an existing browser instance in case of Chromium family browsers.
*/
cdpEndpoint?: string;
/**
* CDP headers to send with the connect request.
*/
cdpHeaders?: Record<string, string>;
/**
* Timeout in milliseconds for connecting to CDP endpoint. Defaults to 30000 (30 seconds). Pass 0 to disable timeout.
*/
cdpTimeout?: number;
/**
* Remote endpoint to connect to an existing Playwright server. May be a
* WebSocket URL string, or a [ConnectOptions] object that mirrors the
* `connectOptions` shape used by the test runner. When passed as an object,
* `exposeNetwork`, `headers`, `slowMo`, and `timeout` are forwarded to the
* underlying connect call.
*/
remoteEndpoint?: string | playwright.ConnectOptions & { endpoint: string };
/**
* Paths to TypeScript files to add as initialization scripts for Playwright page.
*/
initPage?: string[];
/**
* Paths to JavaScript files to add as initialization scripts.
* The scripts will be evaluated in every page before any of the page's scripts.
*/
initScript?: string[];
},
/**
* Connect to a running browser instance (Edge/Chrome only). If specified, `browser`
* config is ignored.
* Requires the "Playwright Extension" to be installed.
*/
extension?: boolean;
server?: {
/**
* The port to listen on for SSE or MCP transport.
*/
port?: number;
/**
* The host to bind the server to. Default is localhost. Use 0.0.0.0 to bind to all interfaces.
*/
host?: string;
/**
* The hosts this server is allowed to serve from. Defaults to the host server is bound to.
* This is not for CORS, but rather for the DNS rebinding protection.
*/
allowedHosts?: string[];
},
/**
* List of enabled tool capabilities. Possible values:
* - 'core': Core browser automation features.
* - 'pdf': PDF generation and manipulation.
* - 'vision': Coordinate-based interactions.
* - 'devtools': Developer tools features.
*/
capabilities?: ToolCapability[];
/**
* Whether to save the Playwright session into the output directory.
*/
saveSession?: boolean;
/**
* Reuse the same browser context between all connected HTTP clients.
*/
sharedBrowserContext?: boolean;
/**
* Secrets are used to replace matching plain text in the tool responses to prevent the LLM
* from accidentally getting sensitive data. It is a convenience and not a security feature,
* make sure to always examine information coming in and from the tool on the client.
*/
secrets?: Record<string, string>;
/**
* The directory to save output files.
*/
outputDir?: string;
/**
* Threshold for evicting old output files, in bytes.
*/
outputMaxSize?: number;
console?: {
/**
* The level of console messages to return. Each level includes the messages of more severe levels. Defaults to "info".
*/
level?: 'error' | 'warning' | 'info' | 'debug';
},
network?: {
/**
* List of origins to allow the browser to request. Default is to allow all. Origins matching both `allowedOrigins` and `blockedOrigins` will be blocked.
*
* Supported formats:
* - Full origin: `https://example.com:8080` - matches only that origin
* - Wildcard port: `http://localhost:*` - matches any port on localhost with http protocol
*/
allowedOrigins?: string[];
/**
* List of origins to block the browser to request. Origins matching both `allowedOrigins` and `blockedOrigins` will be blocked.
*
* Supported formats:
* - Full origin: `https://example.com:8080` - matches only that origin
* - Wildcard port: `http://localhost:*` - matches any port on localhost with http protocol
*/
blockedOrigins?: string[];
};
/**
* Specify the attribute to use for test ids, defaults to "data-testid".
*/
testIdAttribute?: string;
timeouts?: {
/*
* Configures default action timeout: https://playwright.dev/docs/api/class-page#page-set-default-timeout. Defaults to 5000ms.
*/
action?: number;
/*
* Configures default navigation timeout: https://playwright.dev/docs/api/class-page#page-set-default-navigation-timeout. Defaults to 60000ms.
*/
navigation?: number;
/**
* Configures default expect timeout: https://playwright.dev/docs/test-timeouts#expect-timeout. Defaults to 5000ms.
*/
expect?: number;
/**
* How long to wait after each action for triggered work (navigations, requests) to settle before responding. Defaults to 500ms.
*/
settle?: number;
};
/**
* Whether to send image responses to the client. Can be "allow", "omit", or "auto". Defaults to "auto", which sends images if the client can display them.
*/
imageResponses?: 'allow' | 'omit';
snapshot?: {
/**
* When taking snapshots for responses, specifies the mode to use.
*/
mode?: 'full' | 'none';
/**
* Whether to include each element's bounding box as [box=x,y,width,height] in snapshots.
* Coordinates are viewport-relative, in CSS pixels (Element.getBoundingClientRect).
*/
boxes?: boolean;
};
/**
* allowUnrestrictedFileAccess acts as a guardrail to prevent the LLM from accidentally
* wandering outside its intended workspace. It is a convenience defense to catch unintended
* file access, not a secure boundary; a deliberate attempt to reach other directories can be
* easily worked around, so always rely on client-level permissions for true security.
*/
allowUnrestrictedFileAccess?: boolean;
/**
* Specify the language to use for code generation.
*/
codegen?: 'typescript' | 'python' | 'java' | 'csharp' | 'none';
}When running headed browser on system w/o display or from worker processes of the IDEs,
run the MCP server from environment with the DISPLAY and pass the --port flag to enable HTTP transport.
npx @playwright/mcp@latest --port 8931And then in MCP client config, set the url to the HTTP endpoint:
{
"mcpServers": {
"playwright": {
"url": "http://localhost:8931/mcp"
}
}
}Related MCP server: Playwright MCP
Playwright MCP is not a security boundary. See MCP Security Best Practices for guidance on securing your deployment.
NOTE: The Docker implementation only supports headless chromium at the moment.
{
"mcpServers": {
"playwright": {
"command": "docker",
"args": ["run", "-i", "--rm", "--init", "--pull=always", "mcr.microsoft.com/playwright/mcp"]
}
}
}Or If you prefer to run the container as a long-lived service instead of letting the MCP client spawn it, use:
docker run -d -i --rm --init --pull=always \
--entrypoint node \
--name playwright \
-p 8931:8931 \
mcr.microsoft.com/playwright/mcp \
/app/cli.js --headless --browser chromium --no-sandbox --port 8931 --host 0.0.0.0The server will listen on host port 8931 and can be reached by any MCP client.
You can build the Docker image yourself.
docker build -t mcr.microsoft.com/playwright/mcp .import http from 'http';
import { createConnection } from '@playwright/mcp';
import { SSEServerTransport } from '@modelcontextprotocol/sdk/server/sse.js';
http.createServer(async (req, res) => {
// ...
// Creates a headless Playwright MCP server with SSE transport
const connection = await createConnection({ browser: { launchOptions: { headless: true } } });
const transport = new SSEServerTransport('/messages', res);
await connection.connect(transport);
// ...
});browser_click
Title: Click
Description: Perform click on a web page
Parameters:
element (string, optional): Human-readable element description used to obtain permission to interact with the element
target (string): Exact target element reference from the page snapshot, or a unique element selector
doubleClick (boolean, optional): Whether to perform a double click instead of a single click
button (string, optional): Button to click, defaults to left
modifiers (array, optional): Modifier keys to press
Read-only: false
browser_close
Title: Close browser
Description: Close the page
Parameters: None
Read-only: false
browser_console_messages
Title: Get console messages
Description: Returns all console messages
Parameters:
level (string): Level of the console messages to return. Each level includes the messages of more severe levels. Defaults to "info".
all (boolean, optional): Return all console messages since the beginning of the session, not just since the last navigation. Defaults to false.
filename (string, optional): Filename to save the console messages to. If not provided, messages are returned as text.
Read-only: true
browser_drag
Title: Drag mouse
Description: Perform drag and drop between two elements
Parameters:
startElement (string, optional): Human-readable source element description used to obtain the permission to interact with the element
startTarget (string): Exact target element reference from the page snapshot, or a unique element selector
endElement (string, optional): Human-readable target element description used to obtain the permission to interact with the element
endTarget (string): Exact target element reference from the page snapshot, or a unique element selector
Read-only: false
browser_drop
Title: Drop files or data onto an element
Description: Drop files or MIME-typed data onto an element, as if dragged from outside the page. At least one of "paths" or "data" must be provided.
Parameters:
element (string, optional): Human-readable element description used to obtain permission to interact with the element
target (string): Exact target element reference from the page snapshot, or a unique element selector
paths (array, optional): Absolute paths to files to drop onto the element.
data (object, optional): Data to drop, as a map of MIME type to string value (e.g. {"text/plain": "hello", "text/uri-list": "https://example.com"}).
Read-only: false
browser_evaluate
Title: Evaluate JavaScript
Description: Evaluate JavaScript expression on page or element
Parameters:
element (string, optional): Human-readable element description used to obtain permission to interact with the element
target (string, optional): Exact target element reference from the page snapshot, or a unique element selector
function (string): () => { /* code / } or (element) => { / code */ } when element is provided
filename (string, optional): Filename to save the result to. If not provided, result is returned as text.
Read-only: false
browser_file_upload
Title: Upload files
Description: Upload one or multiple files
Parameters:
paths (array, optional): The absolute paths to the files to upload. Can be single file or multiple files. If omitted, file chooser is cancelled.
Read-only: false
browser_fill_form
Title: Fill form
Description: Fill multiple form fields
Parameters:
fields (array): Fields to fill in
Read-only: false
browser_find
Title: Find in page snapshot
Description: Search the accessibility snapshot of the current page for text or a regular expression. Returns matching snapshot nodes with a few lines of surrounding context (like search snippets), each shown under its path from the root of the tree, which is cheaper than capturing the whole snapshot when you only need to locate an element and its ref.
Parameters:
text (string, optional): Plain text to search for in the page snapshot (case-insensitive substring match). Provide either text or regex, not both.
regex (string, optional): Regular expression to search for in the page snapshot. Matching is case-sensitive by default; wrap the pattern in slashes to add flags, e.g. "/error/i" for case-insensitive. Provide either text or regex, not both.
Read-only: true
browser_handle_dialog
Title: Handle a dialog
Description: Handle a dialog
Parameters:
accept (boolean): Whether to accept the dialog.
promptText (string, optional): The text of the prompt in case of a prompt dialog.
Read-only: false
browser_hover
Title: Hover mouse
Description: Hover over element on page
Parameters:
element (string, optional): Human-readable element description used to obtain permission to interact with the element
target (string): Exact target element reference from the page snapshot, or a unique element selector
Read-only: false
browser_navigate
Title: Navigate to a URL
Description: Navigate to a URL
Parameters:
url (string): The URL to navigate to
Read-only: false
browser_navigate_back
Title: Go back
Description: Go back to the previous page in the history
Parameters: None
Read-only: false
browser_network_request
Title: Show network request details
Description: Returns full details (headers and body) of a single network request, or a single part if part is set. Use the number from browser_network_requests.
Parameters:
index (integer): 1-based index of the request, as printed by browser_network_requests.
part (string, optional): Return only this part of the request. Omit to return full details.
filename (string, optional): Filename to save the result to. If not provided, output is returned as text.
Read-only: true
browser_network_requests
Title: List network requests
Description: Returns a numbered list of network requests since loading the page. Use browser_network_request with the number to get full details.
Parameters:
static (boolean): Whether to include successful static resources like images, fonts, scripts, etc. Defaults to false.
filter (string, optional): Only return requests whose URL matches this regexp (e.g. "/api/.*user").
filename (string, optional): Filename to save the network requests to. If not provided, requests are returned as text.
Read-only: true
browser_press_key
Title: Press a key
Description: Press a key on the keyboard
Parameters:
key (string): Name of the key to press or a character to generate, such as ArrowLeft or a
Read-only: false
browser_resize
Title: Resize browser window
Description: Resize the browser window
Parameters:
width (number): Width of the browser window
height (number): Height of the browser window
Read-only: false
browser_run_code_unsafe
Title: Run Playwright code (unsafe)
Description: Run a Playwright code snippet. Unsafe: executes arbitrary JavaScript in the Playwright server process and is RCE-equivalent.
Parameters:
code (string, optional): A JavaScript function containing Playwright code to execute. It will be invoked with a single argument, page, which you can use for any page interaction. For example: async (page) => { await page.getByRole('button', { name: 'Submit' }).click(); return await page.title(); }
filename (string, optional): Load code from the specified file. If both code and filename are provided, code will be ignored.
Read-only: false
browser_select_option
Title: Select option
Description: Select an option in a dropdown
Parameters:
element (string, optional): Human-readable element description used to obtain permission to interact with the element
target (string): Exact target element reference from the page snapshot, or a unique element selector
values (array): Array of values to select in the dropdown. This can be a single value or multiple values.
Read-only: false
browser_snapshot
Title: Page snapshot
Description: Capture accessibility snapshot of the current page, this is better than screenshot
Parameters:
target (string, optional): Exact target element reference from the page snapshot, or a unique element selector
filename (string, optional): Save snapshot to markdown file instead of returning it in the response.
depth (number, optional): Limit the depth of the snapshot tree
boxes (boolean, optional): Include each element's bounding box as [box=x,y,width,height] in the snapshot. Coordinates are viewport-relative, in CSS pixels (Element.getBoundingClientRect)
Read-only: true
browser_take_screenshot
Title: Take a screenshot
Description: Take a screenshot of the current page. You can't perform actions based on the screenshot, use browser_snapshot for actions.
Parameters:
element (string, optional): Human-readable element description used to obtain permission to interact with the element
target (string, optional): Exact target element reference from the page snapshot, or a unique element selector
type (string, optional): Image format for the screenshot. If unset, inferred from the filename extension, otherwise png.
filename (string, optional): File name to save the screenshot to. Defaults to page-{timestamp}.{png|jpeg|webp} if not specified. Prefer relative file names to stay within the output directory.
fullPage (boolean, optional): When true, takes a screenshot of the full scrollable page, instead of the currently visible viewport. Cannot be used with element screenshots.
scale (string): Image resolution scale. "css" produces a screenshot sized in CSS pixels (smaller, consistent across devices). "device" produces a high-resolution screenshot using device pixels (larger, accounts for the device pixel ratio). Default is css.
Read-only: true
browser_type
Title: Type text
Description: Type text into editable element
Parameters:
element (string, optional): Human-readable element description used to obtain permission to interact with the element
target (string): Exact target element reference from the page snapshot, or a unique element selector
text (string): Text to type into the element
submit (boolean, optional): Whether to submit entered text (press Enter after)
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.
Read-only: false
browser_wait_for
Title: Wait for
Description: Wait for text to appear or disappear or a specified time to pass
Parameters:
time (number, optional): The time to wait in seconds
text (string, optional): The text to wait for
textGone (string, optional): The text to wait for to disappear
Read-only: false
browser_tabs
Title: Manage tabs
Description: List, create, close, or select a browser tab.
Parameters:
action (string): Operation to perform
index (number, optional): Tab index, used for close/select. If omitted for close, current tab is closed.
url (string, optional): URL to navigate to in the new tab, used for new.
Read-only: false
browser_get_config
Title: Get config
Description: Get the final resolved config after merging CLI options, environment variables and config file.
Parameters: None
Read-only: true
browser_network_state_set
Title: Set network state
Description: Sets the browser network state to online or offline. When offline, all network requests will fail.
Parameters:
state (string): Set to "offline" to simulate offline mode, "online" to restore network connectivity
Read-only: false
browser_route
Title: Mock network requests
Description: Set up a route to mock network requests matching a URL pattern
Parameters:
pattern (string): URL pattern to match (e.g., "/api/users", "/*.{png,jpg}")
status (number, optional): HTTP status code to return (default: 200)
body (string, optional): Response body (text or JSON string)
contentType (string, optional): Content-Type header (e.g., "application/json", "text/html")
headers (array, optional): Headers to add in "Name: Value" format
removeHeaders (string, optional): Comma-separated list of header names to remove from request
Read-only: false
browser_route_list
Title: List network routes
Description: List all active network routes
Parameters: None
Read-only: true
browser_unroute
Title: Remove network routes
Description: Remove network routes matching a pattern (or all routes if no pattern specified)
Parameters:
pattern (string, optional): URL pattern to unroute (omit to remove all routes)
Read-only: false
browser_cookie_clear
Title: Clear cookies
Description: Clear all cookies
Parameters: None
Read-only: false
browser_cookie_delete
Title: Delete cookie
Description: Delete a specific cookie
Parameters:
name (string): Cookie name to delete
Read-only: false
browser_cookie_get
Title: Get cookie
Description: Get a specific cookie by name
Parameters:
name (string): Cookie name to get
Read-only: true
browser_cookie_list
Title: List cookies
Description: List all cookies (optionally filtered by domain/path)
Parameters:
domain (string, optional): Filter cookies by domain
path (string, optional): Filter cookies by path
Read-only: true
browser_cookie_set
Title: Set cookie
Description: Set a cookie with optional flags (domain, path, expires, httpOnly, secure, sameSite)
Parameters:
name (string): Cookie name
value (string): Cookie value
domain (string, optional): Cookie domain
path (string, optional): Cookie path
expires (number, optional): Cookie expiration as Unix timestamp
httpOnly (boolean, optional): Whether the cookie is HTTP only
secure (boolean, optional): Whether the cookie is secure
sameSite (string, optional): Cookie SameSite attribute
Read-only: false
browser_localstorage_clear
Title: Clear localStorage
Description: Clear all localStorage
Parameters: None
Read-only: false
browser_localstorage_delete
Title: Delete localStorage item
Description: Delete a localStorage item
Parameters:
key (string): Key to delete
Read-only: false
browser_localstorage_get
Title: Get localStorage item
Description: Get a localStorage item by key
Parameters:
key (string): Key to get
Read-only: true
browser_localstorage_list
Title: List localStorage
Description: List all localStorage key-value pairs
Parameters: None
Read-only: true
browser_localstorage_set
Title: Set localStorage item
Description: Set a localStorage item
Parameters:
key (string): Key to set
value (string): Value to set
Read-only: false
browser_sessionstorage_clear
Title: Clear sessionStorage
Description: Clear all sessionStorage
Parameters: None
Read-only: false
browser_sessionstorage_delete
Title: Delete sessionStorage item
Description: Delete a sessionStorage item
Parameters:
key (string): Key to delete
Read-only: false
browser_sessionstorage_get
Title: Get sessionStorage item
Description: Get a sessionStorage item by key
Parameters:
key (string): Key to get
Read-only: true
browser_sessionstorage_list
Title: List sessionStorage
Description: List all sessionStorage key-value pairs
Parameters: None
Read-only: true
browser_sessionstorage_set
Title: Set sessionStorage item
Description: Set a sessionStorage item
Parameters:
key (string): Key to set
value (string): Value to set
Read-only: false
browser_set_storage_state
Title: Restore storage state
Description: Restore storage state (cookies, local storage) from a file. This clears existing cookies and local storage before restoring.
Parameters:
filename (string): Path to the storage state file to restore from
Read-only: false
browser_storage_state
Title: Save storage state
Description: Save storage state (cookies, local storage) to a file for later reuse
Parameters:
filename (string, optional): File name to save the storage state to. Defaults to storage-state-{timestamp}.json if not specified.
Read-only: true
browser_annotate
Title: Annotate the current page
Description: Open the Playwright Dashboard in annotation mode for the current page and wait for the user to draw annotations. Returns the annotated screenshot, ARIA snapshot, and the list of annotations.
Parameters: None
Read-only: true
browser_hide_highlight
Title: Hide element highlight
Description: Remove a highlight overlay previously added for the element.
Parameters:
element (string, optional): Human-readable element description used when adding the highlight; must match the value passed to browser_highlight.
target (string, optional): Exact target element reference from the page snapshot, or a unique element selector
Read-only: true
browser_highlight
Title: Highlight element
Description: Show a persistent highlight overlay around the element on the page.
Parameters:
element (string, optional): Human-readable element description used to obtain permission to interact with the element
target (string): Exact target element reference from the page snapshot, or a unique element selector
style (string, optional): Additional inline CSS applied to the highlight overlay, e.g. "outline: 2px dashed red".
Read-only: true
browser_resume
Title: Resume paused script execution
Description: Resume script execution after it was paused. When called with step set to true, execution will pause again before the next action.
Parameters:
step (boolean, optional): When true, execution will pause again before the next action, allowing step-by-step debugging.
location (string, optional): Pause execution at a specific :, e.g. "example.spec.ts:42".
Read-only: false
browser_start_recording
Title: Start recording user actions
Description: Start recording actions that the user performs in the browser as Playwright code. Use it when the user wants to demonstrate a flow manually. Call browser_stop_recording when the user says they are done to retrieve the recorded actions.
Parameters: None
Read-only: true
browser_start_tracing
Title: Start tracing
Description: Start trace recording
Parameters: None
Read-only: true
browser_start_video
Title: Start video
Description: Start video recording
Parameters:
filename (string, optional): Filename to save the video.
size (object, optional): Video size
Read-only: true
browser_stop_recording
Title: Stop recording user actions
Description: Stop the recording started with browser_start_recording and return the recorded actions as Playwright code.
Parameters: None
Read-only: true
browser_stop_tracing
Title: Stop tracing
Description: Stop trace recording
Parameters: None
Read-only: true
browser_stop_video
Title: Stop video
Description: Stop video recording
Parameters: None
Read-only: true
browser_video_chapter
Title: Video chapter
Description: Add a chapter marker to the video recording. Shows a full-screen chapter card with blurred backdrop.
Parameters:
title (string): Chapter title
description (string, optional): Chapter description
duration (number, optional): Duration in milliseconds to show the chapter card
Read-only: true
browser_video_hide_actions
Title: Hide action overlays
Description: Stop annotating actions performed on the page.
Parameters: None
Read-only: true
browser_video_show_actions
Title: Show action overlays
Description: Annotate subsequent actions performed on the page with a callout that names the action and highlights the target element. Useful while video recording or screencasting.
Parameters:
duration (number, optional): How long each action annotation stays on screen, in milliseconds. Defaults to 500.
position (string, optional): Where to place the action title relative to the page. Defaults to top-right.
cursor (string, optional): Cursor decoration for pointer actions. "pointer" (default) animates a mouse pointer from the previous action point to the next one; "none" disables the cursor decoration.
Read-only: true
browser_mouse_click_xy
Title: Click
Description: Click mouse button at a given position
Parameters:
x (number): X coordinate
y (number): Y coordinate
button (string, optional): Button to click, defaults to left
clickCount (number, optional): Number of clicks, defaults to 1
delay (number, optional): Time to wait between mouse down and mouse up in milliseconds, defaults to 0
Read-only: false
browser_mouse_down
Title: Press mouse down
Description: Press mouse down
Parameters:
button (string, optional): Button to press, defaults to left
Read-only: false
browser_mouse_drag_xy
Title: Drag mouse
Description: Drag left mouse button to a given position
Parameters:
startX (number): Start X coordinate
startY (number): Start Y coordinate
endX (number): End X coordinate
endY (number): End Y coordinate
Read-only: false
browser_mouse_move_xy
Title: Move mouse
Description: Move mouse to a given position
Parameters:
x (number): X coordinate
y (number): Y coordinate
Read-only: false
browser_mouse_up
Title: Press mouse up
Description: Press mouse up
Parameters:
button (string, optional): Button to press, defaults to left
Read-only: false
browser_mouse_wheel
Title: Scroll mouse wheel
Description: Scroll mouse wheel
Parameters:
deltaX (number): X delta
deltaY (number): Y delta
Read-only: false
browser_pdf_save
Title: Save as PDF
Description: Save page as PDF
Parameters:
filename (string, optional): File name to save the pdf to. Defaults to page-{timestamp}.pdf if not specified. Prefer relative file names to stay within the output directory.
Read-only: true
browser_generate_locator
Title: Create locator for element
Description: Generate locator for the given element to use in tests
Parameters:
element (string, optional): Human-readable element description used to obtain permission to interact with the element
target (string): Exact target element reference from the page snapshot, or a unique element selector
Read-only: true
browser_verify_element_visible
Title: Verify element visible
Description: Verify element is visible on the page
Parameters:
role (string): ROLE of the element. Can be found in the snapshot like this: - {ROLE} "Accessible Name":
accessibleName (string): ACCESSIBLE_NAME of the element. Can be found in the snapshot like this: - role "{ACCESSIBLE_NAME}"
Read-only: false
browser_verify_list_visible
Title: Verify list visible
Description: Verify list is visible on the page
Parameters:
element (string): Human-readable list description
target (string): Exact target element reference that points to the list
items (array): Items to verify
Read-only: false
browser_verify_text_visible
Title: Verify text visible
Description: Verify text is visible on the page. Prefer browser_verify_element_visible if possible.
Parameters:
text (string): TEXT to verify. Can be found in the snapshot like this: - role "Accessible Name": {TEXT} or like this: - text: {TEXT}
Read-only: false
browser_verify_value
Title: Verify value
Description: Verify element value
Parameters:
type (string): Type of the element
element (string): Human-readable element description
target (string): Exact target element reference from the page snapshot
value (string): Value to verify. For checkbox, use "true" or "false".
Read-only: false
Perform click on a web page
| Name | Required | Description | Default |
|---|---|---|---|
| button | No | Button to click, defaults to left | |
| target | Yes | Exact target element reference from the page snapshot, or a unique element selector | |
| element | No | Human-readable element description used to obtain permission to interact with the element | |
| modifiers | No | Modifier keys to press | |
| doubleClick | No | Whether to perform a double click instead of a single click |
Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?
Annotations already indicate destructiveHint=true and readOnlyHint=false, so the agent knows this mutates state. However, beyond annotations, the description adds no behavioral context (e.g., whether it waits for navigation, scrolls into view, or handles dialogs). It does not contradict annotations.
Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.
Is the description appropriately sized, front-loaded, and free of redundancy?
The description is extremely short (one sentence), which is concise but arguably too brief. It lacks informative content that could be included without affecting conciseness. It is front-loaded but does not earn its place with substantial value.
Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.
Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?
Given the tool has 5 parameters, no output schema, and many siblings, the description is incomplete. It does not explain the effect of clicking (e.g., navigation, state changes), preconditions, or typical usage patterns. The annotations partially compensate but the description leaves significant gaps.
Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.
Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?
Schema description coverage is 100%, so parameters are well-documented in the schema. The description adds no additional meaning or context beyond what is already provided in the parameter descriptions.
Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.
Does the description clearly state what the tool does and how it differs from similar tools?
The description 'Perform click on a web page' states the action and resource clearly, but does not distinguish from sibling tools like browser_hover or browser_drag. It lacks specificity about element targeting or click types beyond the bare minimum.
Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.
Does the description explain when to use this tool, when not to, or what alternatives exist?
No guidance on when to use this tool versus alternatives. There is no mention of prerequisites (e.g., page must be loaded, element must be visible), exclusions, or scenarios where another tool might be more appropriate.
Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.
Close the page
| Name | Required | Description | Default |
|---|---|---|---|
No parameters | |||
Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?
Annotations already indicate destructiveHint=true, so the destructive nature is clear. However, the description adds no behavioral context beyond that, such as what happens to unsaved data or whether the page can be reopened. With annotations present, the bar is lower, but the description still fails to add value.
Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.
Is the description appropriately sized, front-loaded, and free of redundancy?
The description is extremely concise at 4 words, with no wasted content. For a simple close action, this level of conciseness is optimal and well-structured.
Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.
Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?
Given no parameters and no output schema, the description is barely adequate. It does not explain side effects, such as whether the browser closes the entire window or just the current tab, or how it interacts with multiple tabs. A minimal tool can get by with a short description, but it still lacks completeness.
Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.
Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?
No parameters exist in the schema, and the description does not mention any. Since there are no parameters to explain, the description does not need to add semantic information. Baseline score of 4 is appropriate.
Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.
Does the description clearly state what the tool does and how it differs from similar tools?
The description 'Close the page' is a clear verb+resource combination. It is distinct from sibling tools like browser_navigate or browser_tabs, though very minimal. Could be more specific (e.g., current tab vs page).
Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.
Does the description explain when to use this tool, when not to, or what alternatives exist?
No guidance is provided about when to use this tool versus alternatives like browser_navigate_back or browser_tabs. Agents have no context to decide which tool is appropriate for closing a tab vs page.
Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.
Returns all console messages
| Name | Required | Description | Default |
|---|---|---|---|
| all | No | Return all console messages since the beginning of the session, not just since the last navigation. Defaults to false. | |
| level | Yes | Level of the console messages to return. Each level includes the messages of more severe levels. Defaults to "info". | info |
| filename | No | Filename to save the console messages to. If not provided, messages are returned as text. |
Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?
Annotations already declare readOnlyHint=true and destructiveHint=false. The description adds no additional behavioral context beyond 'returns all console messages', which is consistent but not enriching.
Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.
Is the description appropriately sized, front-loaded, and free of redundancy?
Single sentence that is concise and to the point. No wasted words, and the structure is appropriate for a simple tool.
Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.
Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?
Given the well-described schema and annotations, the description is sufficient for a straightforward read-only tool. It lacks return value details but that is acceptable without an output schema.
Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.
Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?
All three parameters are fully described in the schema (100% coverage). The description adds no extra meaning beyond what the schema already provides, achieving the baseline.
Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.
Does the description clearly state what the tool does and how it differs from similar tools?
Description clearly states it returns console messages, which is a specific verb+resource. However, it does not differentiate from sibling tools like browser_network_requests, but the purpose is unambiguous.
Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.
Does the description explain when to use this tool, when not to, or what alternatives exist?
No guidance on when to use this tool versus alternatives. For example, it doesn't mention that this tool is for accessing client-side console output, distinct from network request logs.
Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.
Perform drag and drop between two elements
| Name | Required | Description | Default |
|---|---|---|---|
| endTarget | Yes | Exact target element reference from the page snapshot, or a unique element selector | |
| endElement | No | Human-readable target element description used to obtain the permission to interact with the element | |
| startTarget | Yes | Exact target element reference from the page snapshot, or a unique element selector | |
| startElement | No | Human-readable source element description used to obtain the permission to interact with the element |
Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?
Annotations already indicate destructiveHint=true and openWorldHint=true, but the description does not elaborate on specific behaviors, such as how drag events are simulated, whether it requires WebDriver or Playwright, or what happens if elements are not draggable. It adds minimal value beyond the annotations.
Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.
Is the description appropriately sized, front-loaded, and free of redundancy?
The description is a single, front-loaded sentence with zero wasted words. It conveys the essential action efficiently.
Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.
Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?
The tool has 4 parameters and no output schema, but the description omits details about return values, error cases, or how the drag-and-drop coordinates are interpreted (e.g., center points vs. specific locations). It fails to fully inform an agent about the tool's capabilities and limitations.
Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.
Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?
The input schema covers all 4 parameters with full description coverage (100%). The tool description does not add any additional meaning to the parameters beyond what the schema already provides, such as clarifying the difference between startElement and startTarget. Baseline of 3 is appropriate given high schema coverage.
Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.
Does the description clearly state what the tool does and how it differs from similar tools?
The description uses a specific verb ('Perform drag and drop') and clearly identifies the resource ('between two elements'). It effectively distinguishes this tool from sibling tools like browser_click, browser_hover, and browser_drop (which might only handle the release phase).
Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.
Does the description explain when to use this tool, when not to, or what alternatives exist?
No guidance is provided on when to use this tool versus alternatives, such as browser_drop for only releasing, or what prerequisites are needed (e.g., obtaining element references via snapshot). The description lacks any contextual hints about ideal scenarios or exclusions.
Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.
Drop files or MIME-typed data onto an element, as if dragged from outside the page. At least one of "paths" or "data" must be provided.
| Name | Required | Description | Default |
|---|---|---|---|
| data | No | Data to drop, as a map of MIME type to string value (e.g. {"text/plain": "hello", "text/uri-list": "https://example.com"}). | |
| paths | No | Absolute paths to files to drop onto the element. | |
| target | Yes | Exact target element reference from the page snapshot, or a unique element selector | |
| element | No | Human-readable element description used to obtain permission to interact with the element |
Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?
Annotations indicate destructiveHint and openWorldHint, so description adds 'as if dragged from outside the page' which clarifies the simulation. However, it does not detail potential side effects like triggering event handlers or page navigation.
Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.
Is the description appropriately sized, front-loaded, and free of redundancy?
Two concise sentences with the core purpose front-loaded. Every word is meaningful and no redundancy.
Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.
Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?
The description is adequate for a simple drop action, but given destructive and open-world annotations, it lacks details on what happens after drop (e.g., event triggering, return value). The schema covers parameters well, but behavioral outcomes are not specified.
Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.
Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?
Schema coverage is 100%, but the description adds a critical constraint not in schema: 'At least one of paths or data must be provided.' This clarifies optionality beyond the required fields.
Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.
Does the description clearly state what the tool does and how it differs from similar tools?
The description clearly states the tool drops files or MIME-typed data onto an element simulating an external drag. It distinguishes from siblings like browser_drag by specifying 'dragged from outside the page' and from browser_file_upload by referring to drop behavior.
Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.
Does the description explain when to use this tool, when not to, or what alternatives exist?
No explicit guidance on when to use this tool over alternatives like browser_drag or browser_file_upload. The hint 'At least one of paths or data must be provided' is a constraint, not a usage guideline.
Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.
Evaluate JavaScript expression on page or element
| Name | Required | Description | Default |
|---|---|---|---|
| target | No | Exact target element reference from the page snapshot, or a unique element selector | |
| element | No | Human-readable element description used to obtain permission to interact with the element | |
| filename | No | Filename to save the result to. If not provided, result is returned as text. | |
| function | Yes | () => { /* code */ } or (element) => { /* code */ } when element is provided |
Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?
Annotations already indicate destructiveHint and openWorldHint, so the description doesn't need to repeat that. However, it doesn't add details about potential side effects or page modifications beyond what annotations provide.
Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.
Is the description appropriately sized, front-loaded, and free of redundancy?
The description is a single, front-loaded sentence with no superfluous words. It effectively conveys the core purpose.
Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.
Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?
While the description covers the basic action, it lacks explanation of return values (e.g., what the JavaScript expression returns) and does not elaborate on the filename parameter for saving results. Given the annotations and schema, it meets minimum viability but has gaps.
Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.
Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?
All parameters are fully described in the schema (100% coverage), so the description adds little extra meaning. The phrase 'on page or element' hints at the element parameter but does not add concrete semantics.
Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.
Does the description clearly state what the tool does and how it differs from similar tools?
The description clearly states it evaluates JavaScript on a page or element, which is a specific verb+resource. However, it does not differentiate from the sibling tool 'browser_run_code_unsafe' which also executes code.
Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.
Does the description explain when to use this tool, when not to, or what alternatives exist?
No guidance is provided on when to use this tool versus alternatives like 'browser_run_code_unsafe' or 'browser_click'. There is no mention of prerequisites, safety considerations, or contextual cues.
Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.
Upload one or multiple files
| Name | Required | Description | Default |
|---|---|---|---|
| paths | No | The absolute paths to the files to upload. Can be single file or multiple files. If omitted, file chooser is cancelled. |
Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?
Annotations include destructiveHint=true and openWorldHint=true, but the description adds no further behavioral context such as side effects (e.g., form submission) or cancellation behavior beyond what the parameter description covers.
Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.
Is the description appropriately sized, front-loaded, and free of redundancy?
Two short, direct sentences with no wasted words, effectively communicating the tool's function.
Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.
Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?
The description is adequate given the simple interface, but could mention the need for a file input element or how the tool interacts with the browser's file chooser for fuller completeness.
Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.
Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?
Schema covers 100% of parameters; the description adds clarity about single vs multiple files and cancellation, complementing the schema's details.
Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.
Does the description clearly state what the tool does and how it differs from similar tools?
The description clearly states the verb 'Upload' and the resource 'one or multiple files', differentiating it from sibling tools like browser_click, browser_navigate, etc.
Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.
Does the description explain when to use this tool, when not to, or what alternatives exist?
No explicit guidance on when to use this tool versus alternatives, nor any prerequisites (e.g., needing a file input element). Implied usage from context.
Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.
Fill multiple form fields
| Name | Required | Description | Default |
|---|---|---|---|
| fields | Yes | Fields to fill in |
Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?
Annotations indicate destructiveHint=true and openWorldHint=true, but the description adds no behavioral context (e.g., whether it overwrites existing values or handles validation). With annotations, the bar is lower, but the description still doesn't supplement.
Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.
Is the description appropriately sized, front-loaded, and free of redundancy?
The description is a single, focused sentence with no irrelevant information, achieving maximal conciseness.
Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.
Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?
While the schema covers parameters, the description lacks context about behavior (e.g., clearing fields, submitting), and there is no output schema. It is minimally complete given the tool's complexity.
Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.
Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?
Schema description coverage is 100%, so the input schema fully documents parameters. The description adds no extra semantics beyond what the schema provides, hence baseline 3.
Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.
Does the description clearly state what the tool does and how it differs from similar tools?
The description 'Fill multiple form fields' uses a specific verb (fill) and resource (form fields), clearly distinguishing it from siblings like browser_type (types into a single field) and browser_click.
Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.
Does the description explain when to use this tool, when not to, or what alternatives exist?
The description implies usage for filling forms but provides no explicit when-to-use or when-not-to-use guidance, nor does it mention alternatives like browser_type for individual fields.
Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.
Search the accessibility snapshot of the current page for text or a regular expression. Returns matching snapshot nodes with a few lines of surrounding context (like search snippets), each shown under its path from the root of the tree, which is cheaper than capturing the whole snapshot when you only need to locate an element and its ref.
| Name | Required | Description | Default |
|---|---|---|---|
| text | No | Plain text to search for in the page snapshot (case-insensitive substring match). Provide either text or regex, not both. | |
| regex | No | Regular expression to search for in the page snapshot. Matching is case-sensitive by default; wrap the pattern in slashes to add flags, e.g. "/error/i" for case-insensitive. Provide either text or regex, not both. |
Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?
Annotations already declare readOnlyHint=true and destructiveHint=false. The description adds that the search is on the accessibility snapshot, returns nodes with surrounding context and paths, and is cheaper than a snapshot. This provides useful behavioral context beyond annotations without contradiction.
Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.
Is the description appropriately sized, front-loaded, and free of redundancy?
The description is two sentences, front-loaded with the main purpose, and contains no redundant information. Every part is meaningful and efficiently conveys the tool's function and usage.
Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.
Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?
Given the tool's simplicity, absence of output schema, and good annotations, the description covers all needed context: what it does, what the return looks like, and when it's appropriate. It is complete for effective selection and invocation.
Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.
Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?
Schema coverage is 100% with descriptions for both parameters. The description adds extra semantics: case-insensitive substring for text, case-sensitive by default for regex with flag syntax. This enhances understanding beyond the schema fields.
Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.
Does the description clearly state what the tool does and how it differs from similar tools?
The description clearly states it searches the accessibility snapshot for text or regex and returns matching nodes with context. It distinguishes itself from 'browser_snapshot' by noting it's cheaper when only locating an element, providing a specific verb and resource.
Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.
Does the description explain when to use this tool, when not to, or what alternatives exist?
The description explains when to use this tool (to locate an element cheaply) and provides constraints ('provide either text or regex, not both'). It implies when not to use (when full snapshot is needed) but does not explicitly list alternatives beyond the sibling comparison.
Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.
Handle a dialog
| Name | Required | Description | Default |
|---|---|---|---|
| accept | Yes | Whether to accept the dialog. | |
| promptText | No | The text of the prompt in case of a prompt dialog. |
Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?
Annotations indicate destructiveHint=true and readOnlyHint=false, but the description adds no additional behavioral context. It does not explain that the tool requires an active dialog, what happens to the dialog after handling, or that prompt dialogs may need text input beyond the schema.
Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.
Is the description appropriately sized, front-loaded, and free of redundancy?
The description is very concise (one sentence), but it omits important context. While not verbose, it is under-specified, earning a middle score.
Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.
Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?
Given the tool's role in browser automation and the presence of many sibling tools, the description fails to explain its context (e.g., only works when a dialog is open) or how it fits into a workflow. No output schema exists, but the description does not compensate by describing return behavior.
Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.
Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?
Schema description coverage is 100%, with clear descriptions for 'accept' and 'promptText'. The description adds no extra meaning but does not need to, as the schema is sufficient. Baseline score of 3 applies.
Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.
Does the description clearly state what the tool does and how it differs from similar tools?
The description 'Handle a dialog' is slightly more specific than the name, indicating the tool deals with browser dialogs. However, 'handle' is a generic verb and does not specify whether it accepts, dismisses, or inputs text, leaving the purpose somewhat vague.
Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.
Does the description explain when to use this tool, when not to, or what alternatives exist?
No guidance is provided about when to use this tool, such as after a dialog appears or as an alternative to other browser tools. The description lacks any contextual cues for selecting it over siblings like browser_click or browser_navigate.
Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.
Hover over element on page
| Name | Required | Description | Default |
|---|---|---|---|
| target | Yes | Exact target element reference from the page snapshot, or a unique element selector | |
| element | No | Human-readable element description used to obtain permission to interact with the element |
Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?
The description provides no behavioral details beyond what annotations indicate. Annotations show destructiveHint=true and openWorldHint=true, but the description does not explain what effects hovering has on the page state or interaction flow.
Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.
Is the description appropriately sized, front-loaded, and free of redundancy?
The description is a single, concise sentence with no wasted words. It is appropriately front-loaded but could potentially include a bit more detail without losing conciseness.
Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.
Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?
Given the tool's simplicity and lack of output schema, the description is minimally adequate. However, it omits mention of potential side effects, timing, or return behavior, which are relevant given the openWorld and destructive hints.
Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.
Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?
The description of the 'element' parameter adds meaningful context about its role in obtaining permission, and the 'target' parameter description clarifies acceptable selectors. With 100% schema description coverage, the description enhances understanding beyond the schema alone.
Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.
Does the description clearly state what the tool does and how it differs from similar tools?
The description 'Hover over element on page' clearly states the verb and resource, making the tool's purpose immediately understandable. However, it does not explicitly distinguish itself from sibling tools like browser_click or browser_drag, which could provide nuance.
Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.
Does the description explain when to use this tool, when not to, or what alternatives exist?
No guidance is given on when to use this tool versus alternatives such as browser_click or browser_type. There is no mention of prerequisites or context where hovering is appropriate.
Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.
Returns full details (headers and body) of a single network request, or a single part if part is set. Use the number from browser_network_requests.
| Name | Required | Description | Default |
|---|---|---|---|
| part | No | Return only this part of the request. Omit to return full details. | |
| index | Yes | 1-based index of the request, as printed by browser_network_requests. | |
| filename | No | Filename to save the result to. If not provided, output is returned as text. |
Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?
Annotations indicate read-only and safe; description adds details about returning full details or selective parts. Does not contradict annotations.
Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.
Is the description appropriately sized, front-loaded, and free of redundancy?
Two sentences, front-loaded with purpose, no wasted words.
Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.
Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?
No output schema, but description covers return format (headers, body, or part) and file saving. Sufficient for a simple retrieval tool.
Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.
Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?
Schema coverage is 100%, so baseline 3. Description adds context for 'part' (omit for full) and 'filename' (save vs text output), enhancing beyond schema.
Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.
Does the description clearly state what the tool does and how it differs from similar tools?
Clearly states it returns details of a single network request, distinguishes from 'browser_network_requests' which lists them. Specific verb+resource.
Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.
Does the description explain when to use this tool, when not to, or what alternatives exist?
Explicitly instructs to use the number from browser_network_requests. No explicit when-not-to-use, but context is clear.
Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.
Returns a numbered list of network requests since loading the page. Use browser_network_request with the number to get full details.
| Name | Required | Description | Default |
|---|---|---|---|
| filter | No | Only return requests whose URL matches this regexp (e.g. "/api/.*user"). | |
| static | Yes | Whether to include successful static resources like images, fonts, scripts, etc. Defaults to false. | |
| filename | No | Filename to save the network requests to. If not provided, requests are returned as text. |
Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?
Annotations already indicate readOnlyHint=true, description adds scope (since page load) and output format (numbered list), but no other behavioral traits.
Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.
Is the description appropriately sized, front-loaded, and free of redundancy?
Two sentences, no redundancy, front-loaded with purpose.
Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.
Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?
No output schema but description hints at output format; lacks pagination/limit info but adequate for a read-only list tool.
Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.
Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?
Schema covers all parameters with descriptions; description adds no extra semantic value beyond what schema provides.
Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.
Does the description clearly state what the tool does and how it differs from similar tools?
Clearly states it returns a numbered list of network requests since page load, and distinguishes from sibling browser_network_request for full details.
Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.
Does the description explain when to use this tool, when not to, or what alternatives exist?
Explicitly mentions when to use sibling for full details, but lacks constraints or when-not-to-use scenarios.
Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.
Press a key on the keyboard
| Name | Required | Description | Default |
|---|---|---|---|
| key | Yes | Name of the key to press or a character to generate, such as `ArrowLeft` or `a` |
Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?
Annotations already indicate destructiveHint=true and readOnlyHint=false, so the description need not repeat that. However, it adds no extra context about keyboard events or focus requirements, which is acceptable given the tool's simplicity.
Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.
Is the description appropriately sized, front-loaded, and free of redundancy?
The description is extremely concise (5 words) with no wasted text. It could benefit from slight expansion for clarity, but its brevity aligns with the tool's simplicity.
Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.
Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?
Given the single parameter, no output schema, and annotations covering safety, the description is adequate for a basic key press action. However, it doesn't mention potential effects like keydown/keyup events or prerequisite focus, which limits completeness.
Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.
Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?
Schema coverage is 100% with a clear parameter description. The tool's description adds no additional semantic value beyond what the schema already provides, so a baseline score of 3 is appropriate.
Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.
Does the description clearly state what the tool does and how it differs from similar tools?
The description 'Press a key on the keyboard' explicitly states the action and resource, clearly distinguishing it from sibling tools like browser_type (for typing text) and browser_click (for clicking).
Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.
Does the description explain when to use this tool, when not to, or what alternatives exist?
No guidance is provided on when to use this tool versus alternatives (e.g., browser_type for character strings, browser_hover for mouse actions). The description lacks context for appropriate usage.
Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.
Resize the browser window
| Name | Required | Description | Default |
|---|---|---|---|
| width | Yes | Width of the browser window | |
| height | Yes | Height of the browser window |
Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?
Annotations already mark destructiveHint: true and readOnlyHint: false, indicating state change. The description adds no additional behavioral context (e.g., effect on layout, timing).
Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.
Is the description appropriately sized, front-loaded, and free of redundancy?
Single sentence, no extraneous words. Appropriate length for a simple tool.
Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.
Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?
For a simple resize operation, the description is adequate but omits units (e.g., pixels) and validation limits. With no output schema, more detail would help.
Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.
Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?
Schema coverage is 100%, with clear descriptions for width and height. The tool description adds no extra semantic meaning beyond what the schema provides.
Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.
Does the description clearly state what the tool does and how it differs from similar tools?
The description 'Resize the browser window' is a specific verb-resource pair. It clearly distinguishes from sibling tools like browser_navigate or browser_click, which have different actions.
Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.
Does the description explain when to use this tool, when not to, or what alternatives exist?
No guidance on when to use this tool versus alternatives. The description does not mention context, prerequisites, or when not to use it.
Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.
Run a Playwright code snippet. Unsafe: executes arbitrary JavaScript in the Playwright server process and is RCE-equivalent.
| Name | Required | Description | Default |
|---|---|---|---|
| code | No | A JavaScript function containing Playwright code to execute. It will be invoked with a single argument, page, which you can use for any page interaction. For example: `async (page) => { await page.getByRole('button', { name: 'Submit' }).click(); return await page.title(); }` | |
| filename | No | Load code from the specified file. If both code and filename are provided, code will be ignored. |
Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?
Annotations already provide destructiveHint=true and openWorldHint=true. The description reinforces this by labeling it 'Unsafe' and 'RCE-equivalent', but does not add new behavioral details beyond what annotations convey. It is consistent with annotations.
Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.
Is the description appropriately sized, front-loaded, and free of redundancy?
The description is a single, front-loaded sentence that immediately states the action and critical warning. No words are wasted; every part earns its place.
Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.
Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?
Given the tool's power and lack of output schema, the description could mention return values or error handling. However, for a code runner, the return is implicit in the code. The warning is sufficient, but completeness is adequate.
Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.
Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?
Schema description coverage is 100%, with parameters well-described in the schema itself (including a full example for 'code'). The tool description adds no extra parameter semantics beyond the schema, so baseline of 3 applies.
Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.
Does the description clearly state what the tool does and how it differs from similar tools?
The description clearly states the verb 'Run' and the resource 'Playwright code snippet'. The warning about unsafety and RCE-equivalency differentiates it from sibling tools like browser_evaluate, making its purpose distinct and specific.
Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.
Does the description explain when to use this tool, when not to, or what alternatives exist?
The description implies through its warning that this tool is for advanced use cases requiring server-side code execution, but it does not explicitly state when to use it versus alternatives like browser_evaluate. No exclusions or when-not guidance is provided.
Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.
Select an option in a dropdown
| Name | Required | Description | Default |
|---|---|---|---|
| target | Yes | Exact target element reference from the page snapshot, or a unique element selector | |
| values | Yes | Array of values to select in the dropdown. This can be a single value or multiple values. | |
| element | No | Human-readable element description used to obtain permission to interact with the element |
Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?
Annotations provide destructiveHint=true, readOnlyHint=false, openWorldHint=true, but the description adds no further behavioral context (e.g., side effects like triggering change events, need for the dropdown to be expanded, or permission flow beyond element description). The description does not contradict annotations.
Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.
Is the description appropriately sized, front-loaded, and free of redundancy?
A single, front-loaded sentence that is concise and contains no redundant information. Perfectly sized for this simple tool.
Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.
Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?
The tool is simple with well-documented schema, but the description does not clarify that 'values' refers to option value attributes or that the element must be a <select>. Missing minor context that would help an agent.
Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.
Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?
Input schema coverage is 100% with descriptions for all three parameters. The description adds no extra meaning beyond what the schema already provides, so the baseline score of 3 is appropriate.
Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.
Does the description clearly state what the tool does and how it differs from similar tools?
The description 'Select an option in a dropdown' uses a specific verb ('select') and clearly identifies the resource (dropdown option). This effectively distinguishes it from sibling tools like browser_click, browser_type, and browser_fill_form.
Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.
Does the description explain when to use this tool, when not to, or what alternatives exist?
No guidance is provided on when to use this tool versus alternatives (e.g., browser_fill_form for filling dropdowns via text, or browser_type for interacting with non-select elements). The description lacks any contextual usage advice.
Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.
Capture accessibility snapshot of the current page, this is better than screenshot
| Name | Required | Description | Default |
|---|---|---|---|
| boxes | No | Include each element's bounding box as [box=x,y,width,height] in the snapshot. Coordinates are viewport-relative, in CSS pixels (Element.getBoundingClientRect) | |
| depth | No | Limit the depth of the snapshot tree | |
| target | No | Exact target element reference from the page snapshot, or a unique element selector | |
| filename | No | Save snapshot to markdown file instead of returning it in the response. |
Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?
Annotations already indicate readOnlyHint=true and non-destructive. The description adds 'accessibility snapshot' context but doesn't elaborate on behavioral traits like what happens to the page state.
Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.
Is the description appropriately sized, front-loaded, and free of redundancy?
The description is a single sentence, concise and front-loaded. It could benefit from slightly more structure but is effective and not verbose.
Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.
Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?
Given four parameters and no output schema, the description is too brief. It fails to explain the return format, the effect of parameters like depth or boxes, or how the snapshot differs from a screenshot.
Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.
Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?
Schema coverage is 100%, so each parameter has a description. The tool description does not add additional meaning beyond the schema, so baseline 3 is appropriate.
Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.
Does the description clearly state what the tool does and how it differs from similar tools?
The description clearly states the tool captures an accessibility snapshot and claims it's better than a screenshot, distinguishing it from sibling tools like browser_take_screenshot.
Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.
Does the description explain when to use this tool, when not to, or what alternatives exist?
The description implies use for accessibility over screenshot, but lacks explicit guidance on when to use vs alternatives like browser_evaluate or when not to use.
Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.
List, create, close, or select a browser tab.
| Name | Required | Description | Default |
|---|---|---|---|
| url | No | URL to navigate to in the new tab, used for new. | |
| index | No | Tab index, used for close/select. If omitted for close, current tab is closed. | |
| action | Yes | Operation to perform |
Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?
Annotations already indicate destructiveHint=true and readOnlyHint=false, which the description aligns with. The description adds minimal behavioral detail beyond the schema, such as 'if omitted for close, current tab is closed,' but does not elaborate on consequences of create or select.
Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.
Is the description appropriately sized, front-loaded, and free of redundancy?
The description is a single, clear sentence covering all intended actions. It is efficient but lacks structural elements like bullet points or separate lines for each action, though not necessary given brevity.
Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.
Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?
For a tab management tool with multiple actions and no output schema, the description covers the core functionality but omits details like what happens when listing empty tabs, the effect of 'select' on focus, or error states. It is minimally adequate.
Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.
Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?
Input schema has 100% coverage with descriptions for action, index, and url. The description only restates the actions and does not add new semantics or examples for parameter usage.
Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.
Does the description clearly state what the tool does and how it differs from similar tools?
The description explicitly states the four actions (list, create, close, select) and the resource (browser tab). It clearly distinguishes from sibling tools like browser_navigate (which operates within a tab) and browser_close (which likely closes the browser, not a tab).
Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.
Does the description explain when to use this tool, when not to, or what alternatives exist?
The description implies usage for basic tab management but provides no explicit guidance on when to use this tool versus alternatives like browser_navigate or browser_close. It does not mention prerequisites or when not to use it.
Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.
Take a screenshot of the current page. You can't perform actions based on the screenshot, use browser_snapshot for actions.
| Name | Required | Description | Default |
|---|---|---|---|
| type | No | Image format for the screenshot. If unset, inferred from the filename extension, otherwise png. | |
| scale | Yes | Image resolution scale. "css" produces a screenshot sized in CSS pixels (smaller, consistent across devices). "device" produces a high-resolution screenshot using device pixels (larger, accounts for the device pixel ratio). Default is css. | css |
| target | No | Exact target element reference from the page snapshot, or a unique element selector | |
| element | No | Human-readable element description used to obtain permission to interact with the element | |
| filename | No | File name to save the screenshot to. Defaults to `page-{timestamp}.{png|jpeg|webp}` if not specified. Prefer relative file names to stay within the output directory. | |
| fullPage | No | When true, takes a screenshot of the full scrollable page, instead of the currently visible viewport. Cannot be used with element screenshots. |
Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?
Annotations already mark this as read-only and non-destructive. The description adds an important behavioral limitation: the screenshot cannot be used as a basis for actions, which is not captured in annotations. However, it doesn't describe how the screenshot is returned or any side effects beyond the annotation hints.
Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.
Is the description appropriately sized, front-loaded, and free of redundancy?
Two concise sentences: the first states the purpose, the second provides a crucial usage caveat. No redundant words, and the structure effectively front-loads the core action before delivering the warning.
Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.
Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?
While the schema covers parameters thoroughly, the description omits what the tool returns (e.g., file path, image data) after taking the screenshot. The note about not using it for actions suggests the image is for visual inspection, but this isn't explicit. For a tool with no output schema, this is a notable gap.
Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.
Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?
All 6 parameters are fully documented in the schema with detailed descriptions (e.g., scale explains CSS vs device pixels, fullPage notes it cannot be combined with element). The description itself adds no parameter-level detail, so it aligns with the schema-heavy baseline of 3.
Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.
Does the description clearly state what the tool does and how it differs from similar tools?
States 'Take a screenshot of the current page' with a specific verb and resource. It distinguishes itself from browser_snapshot by explicitly noting 'use browser_snapshot for actions,' clarifying that this tool is for visual capture while the alternative is for actions.
Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.
Does the description explain when to use this tool, when not to, or what alternatives exist?
Explicitly warns 'You can't perform actions based on the screenshot, use browser_snapshot for actions,' providing a clear exclusion and naming the alternative. This gives the agent concrete guidance on when not to use this tool and which sibling to use instead.
Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.
Type text into editable element
| Name | Required | Description | Default |
|---|---|---|---|
| text | Yes | Text to type into the element | |
| slowly | No | 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 | No | Whether to submit entered text (press Enter after) | |
| target | Yes | Exact target element reference from the page snapshot, or a unique element selector | |
| element | No | Human-readable element description used to obtain permission to interact with the element |
Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?
Annotations already declare destructiveHint=true (modification). Description adds minimal behavioral context; does not mention the submit or slowly parameters that affect behavior.
Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.
Is the description appropriately sized, front-loaded, and free of redundancy?
Single sentence, no fluff. Front-loaded with verb and object. Efficient use of space.
Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.
Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?
With 5 parameters and no output schema, the description is too brief. It omits details like target specification, editable element requirements, and return behavior.
Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.
Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?
All parameters are documented in the input schema (100% coverage). The description does not add additional meaning beyond what the schema provides, baseline score of 3.
Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.
Does the description clearly state what the tool does and how it differs from similar tools?
The description 'Type text into editable element' clearly identifies the verb (type) and resource (editable element), distinguishing it from sibling tools like browser_click or browser_fill_form.
Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.
Does the description explain when to use this tool, when not to, or what alternatives exist?
No guidance on when to use this tool versus alternatives (e.g., browser_fill_form, browser_press_key). Missing context about prerequisites or when not to use.
Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.
Wait for text to appear or disappear or a specified time to pass
| Name | Required | Description | Default |
|---|---|---|---|
| text | No | The text to wait for | |
| time | No | The time to wait in seconds | |
| textGone | No | The text to wait for to disappear |
Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?
Annotations mark the tool as read-only and non-destructive, which the description reinforces. However, the description does not disclose behavior when multiple parameters are combined (e.g., both text and time), potential error conditions, or timeout behavior, which are important for correct invocation.
Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.
Is the description appropriately sized, front-loaded, and free of redundancy?
The description is a single, front-loaded sentence that efficiently conveys the tool's core functionality with no extraneous information.
Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.
Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?
Given no output schema and simple behavior, the description lacks critical details about parameter combination logic, error handling, and return behavior. Agents may need to infer behavior from the parameter descriptions alone, which is insufficient for robust decision-making.
Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.
Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?
Schema coverage is 100% with each parameter described. The description adds context by linking 'text' to 'appear', 'textGone' to 'disappear', and 'time' to a wait duration. However, it does not explain interaction between parameters or provide formatting details beyond the schema.
Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.
Does the description clearly state what the tool does and how it differs from similar tools?
The description clearly states the tool's purpose: waiting for text to appear, disappear, or a specified time. It distinguishes itself from sibling browser action tools like click or navigate by focusing on waiting.
Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.
Does the description explain when to use this tool, when not to, or what alternatives exist?
The description implies usage scenarios (waiting for text or time) but does not explicitly state when to use this tool versus alternatives like polling or other wait mechanisms. No exclusions or when-not-to-use guidance is provided.
Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.
Most tools have distinct purposes, but there is potential confusion between browser_evaluate and browser_run_code_unsafe (both execute JS, but one is RCE-equivalent) and between browser_snapshot and browser_take_screenshot (descriptions clarify but agent might still misselect). Overall, the tools are well-differentiated.
All tools start with 'browser_', but the naming pattern is inconsistent: some are verbs (browser_click), some verb_noun (browser_fill_form), and some nouns (browser_tabs, browser_network_requests). The mix of styles reduces predictability.
With 24 tools, the set covers a wide range of browser automation actions without being overwhelmingly large. Each tool serves a clear purpose, though a few could be consolidated (e.g., browser_network_request and browser_network_requests).
The tool surface covers most common browser automation tasks: navigation, input, form filling, file upload, network, console, dialogs, tabs, and screenshots. Minor gaps exist (e.g., cookie management, frame handling), but the core workflows are well-supported.
Unclaimed servers have limited discoverability.
If you are the server author, to access and configure the admin panel.
A comprehensive Model Context Protocol (MCP) server that enables AI assistants to interact with yo…
E2LLM gives your AI eyes and hands in a real browser: structured perception (SiFR) plus action.
A Model Context Protocol server for Wix AI tools
Enable secure connectivity between Sentry issues and debugging data, and LLM clients, using a Model Context Protocol (MCP) server.
We provide all the information about MCP servers via our MCP API.
curl -X GET 'https://glama.ai/api/mcp/v1/servers/microsoft/playwright-mcp'
If you have feedback or need assistance with the MCP directory API, please join our Discord server
