Skip to main content
Glama
lewisvoncken

Playwright MCP

by lewisvoncken

browser_take_screenshot

Read-only

Capture screenshots of web pages or specific elements using Playwright MCP's browser automation. Save images in PNG or JPEG format for documentation, testing, or visual reference purposes.

Instructions

Take a screenshot of the current page. You can't perform actions based on the screenshot, use browser_snapshot for actions.

Input Schema

TableJSON Schema
NameRequiredDescriptionDefault
typeNoImage format for the screenshot. Default is png.png
filenameNoFile name to save the screenshot to. Defaults to `page-{timestamp}.{png|jpeg}` if not specified. Prefer relative file names to stay within the output directory.
elementNoHuman-readable element description used to obtain permission to screenshot the element. If not provided, the screenshot will be taken of viewport. If element is provided, ref must be provided too.
refNoExact target element reference from the page snapshot. If not provided, the screenshot will be taken of viewport. If ref is provided, element must be provided too.
fullPageNoWhen true, takes a screenshot of the full scrollable page, instead of the currently visible viewport. Cannot be used with element screenshots.

Implementation Reference

  • The main handler function for the 'browser_take_screenshot' tool. It captures a screenshot of the current page viewport or a specific element using Playwright, handles options like format (PNG/JPEG), quality, saves to a file, and returns base64-encoded image content if supported by the client.
    handle: async (context, params) => {
      const tab = context.currentTabOrDie();
      const snapshot = tab.snapshotOrDie();
      const fileType = params.raw ? 'png' : 'jpeg';
      const fileName = await outputFile(context.config, params.filename ?? `page-${new Date().toISOString()}.${fileType}`);
      const options: playwright.PageScreenshotOptions = { type: fileType, quality: fileType === 'png' ? undefined : 50, scale: 'css', path: fileName };
      const isElementScreenshot = params.element && params.ref;
    
      const code = [
        `// Screenshot ${isElementScreenshot ? params.element : 'viewport'} and save it as ${fileName}`,
      ];
    
      const locator = params.ref ? snapshot.refLocator({ element: params.element || '', ref: params.ref }) : null;
    
      if (locator)
        code.push(`await page.${await generateLocator(locator)}.screenshot(${javascript.formatObject(options)});`);
      else
        code.push(`await page.screenshot(${javascript.formatObject(options)});`);
    
      const includeBase64 = context.clientSupportsImages();
      const action = async () => {
        const screenshot = locator ? await locator.screenshot(options) : await tab.page.screenshot(options);
        return {
          content: includeBase64 ? [{
            type: 'image' as 'image',
            data: screenshot.toString('base64'),
            mimeType: fileType === 'png' ? 'image/png' : 'image/jpeg',
          }] : []
        };
      };
    
      return {
        code,
        action,
        captureSnapshot: true,
        waitForNetwork: false,
      };
    }
  • Zod schema defining the input parameters for the tool: raw (boolean for PNG without compression), filename (optional string), element and ref (paired strings for element screenshot with validation refinement).
    const screenshotSchema = z.object({
      raw: z.boolean().optional().describe('Whether to return without compression (in PNG format). Default is false, which returns a JPEG image.'),
      filename: z.string().optional().describe('File name to save the screenshot to. Defaults to `page-{timestamp}.{png|jpeg}` if not specified.'),
      element: z.string().optional().describe('Human-readable element description used to obtain permission to screenshot the element. If not provided, the screenshot will be taken of viewport. If element is provided, ref must be provided too.'),
      ref: z.string().optional().describe('Exact target element reference from the page snapshot. If not provided, the screenshot will be taken of viewport. If ref is provided, element must be provided too.'),
    }).refine(data => {
      return !!data.element === !!data.ref;
    }, {
      message: 'Both element and ref must be provided or neither.',
      path: ['ref', 'element']
    });
  • Tool registration via defineTool call, which combines the schema, name 'browser_take_screenshot', description, and handler into a tool object, exported for inclusion in the tools list.
    const screenshot = defineTool({
      capability: 'core',
      schema: {
        name: '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.`,
        inputSchema: screenshotSchema,
        type: 'readOnly',
      },
    
      handle: async (context, params) => {
        const tab = context.currentTabOrDie();
        const snapshot = tab.snapshotOrDie();
        const fileType = params.raw ? 'png' : 'jpeg';
        const fileName = await outputFile(context.config, params.filename ?? `page-${new Date().toISOString()}.${fileType}`);
        const options: playwright.PageScreenshotOptions = { type: fileType, quality: fileType === 'png' ? undefined : 50, scale: 'css', path: fileName };
        const isElementScreenshot = params.element && params.ref;
    
        const code = [
          `// Screenshot ${isElementScreenshot ? params.element : 'viewport'} and save it as ${fileName}`,
        ];
    
        const locator = params.ref ? snapshot.refLocator({ element: params.element || '', ref: params.ref }) : null;
    
        if (locator)
          code.push(`await page.${await generateLocator(locator)}.screenshot(${javascript.formatObject(options)});`);
        else
          code.push(`await page.screenshot(${javascript.formatObject(options)});`);
    
        const includeBase64 = context.clientSupportsImages();
        const action = async () => {
          const screenshot = locator ? await locator.screenshot(options) : await tab.page.screenshot(options);
          return {
            content: includeBase64 ? [{
              type: 'image' as 'image',
              data: screenshot.toString('base64'),
              mimeType: fileType === 'png' ? 'image/png' : 'image/jpeg',
            }] : []
          };
        };
    
        return {
          code,
          action,
          captureSnapshot: true,
          waitForNetwork: false,
        };
      }
    });
  • src/tools.ts:36-52 (registration)
    The screenshot tool is included in the snapshotTools array (spread from its module export), which serves as the primary list of available tools for the MCP server.
    export const snapshotTools: Tool<any>[] = [
      ...common(true),
      ...console,
      ...dialogs(true),
      ...files(true),
      ...install,
      ...keyboard(true),
      ...navigate(true),
      ...network,
      ...pdf,
      ...screenshot,
      ...snapshot,
      ...tabs(true),
      ...testing,
      ...video,
      ...wait(true),
    ];
  • Final registration to MCP server: snapshotTools selected as allTools based on config, filtered by capabilities, and handled in ListTools and CallTool request handlers.
    export function createConnection(config: FullConfig, browserContextFactory: BrowserContextFactory): Connection {
      const allTools = config.vision ? visionTools : snapshotTools;
      const tools = allTools.filter(tool => !config.capabilities || tool.capability === 'core' || config.capabilities.includes(tool.capability));
      validateConfig(config);
      const context = new Context(tools, config, browserContextFactory);
Behavior3/5

Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?

Annotations already declare readOnlyHint=true, destructiveHint=false, and openWorldHint=true, covering the safety profile. The description adds useful context about the inability to perform actions with the screenshot, which isn't captured in annotations. However, it doesn't describe behavioral aspects like what happens with permissions for element screenshots or how the tool handles errors.

Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.

Conciseness5/5

Is the description appropriately sized, front-loaded, and free of redundancy?

Two sentences with zero waste. The first sentence states the core purpose, the second provides crucial usage guidance. Every word earns its place, and the structure is front-loaded with the most important information.

Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.

Completeness4/5

Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?

For a read-only tool with good annotations and full schema coverage, the description provides adequate context. It covers purpose and sibling differentiation well. The main gap is the lack of output information (no output schema and description doesn't mention what's returned), but given the tool's relative simplicity, this is a minor omission.

Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.

Parameters3/5

Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?

Schema description coverage is 100%, so the schema already documents all 5 parameters thoroughly. The description doesn't add any parameter-specific information beyond what's in the schema, but the schema provides complete coverage, justifying the baseline score.

Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.

Purpose5/5

Does the description clearly state what the tool does and how it differs from similar tools?

The description clearly states the specific action ('Take a screenshot') and resource ('current page'), distinguishing it from sibling tools like browser_snapshot which is mentioned for actions. It provides a complete verb+resource+scope statement.

Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.

Usage Guidelines5/5

Does the description explain when to use this tool, when not to, or what alternatives exist?

Explicitly provides when-not-to-use guidance by stating 'You can't perform actions based on the screenshot, use browser_snapshot for actions.' This clearly distinguishes this tool from its sibling and provides clear alternative usage context.

Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.

Install Server

Other Tools

Latest Blog Posts

MCP directory API

We provide all the information about MCP servers via our MCP API.

curl -X GET 'https://glama.ai/api/mcp/v1/servers/lewisvoncken/playwright-mcp'

If you have feedback or need assistance with the MCP directory API, please join our Discord server