screenshot

Instructions

Implementation Reference

• index-browser-only.js:350-380 (handler)

  Main handler function for the 'screenshot' tool. Uses CDP to capture screenshot (full page optional) and saves PNG to /tmp/{name}.

  async screenshot(name = 'screenshot.png', fullPage = false) {
    await this.ensureChromium();
    
    const screenshotParams = { format: 'png' };
    
    if (fullPage) {
      const metrics = await this.sendCDPCommand('Page.getLayoutMetrics');
      await this.sendCDPCommand('Emulation.setDeviceMetricsOverride', {
        width: metrics.contentSize.width,
        height: metrics.contentSize.height,
        deviceScaleFactor: 1,
        mobile: false,
      });
      screenshotParams.clip = {
        x: 0,
        y: 0,
        width: metrics.contentSize.width,
        height: metrics.contentSize.height,
        scale: 1,
      };
    }
  
    const screenshot = await this.sendCDPCommand('Page.captureScreenshot', screenshotParams);
    const screenshotPath = `/tmp/${name}`;
    
    fs.writeFileSync(screenshotPath, screenshot.data, 'base64');
    
    return {
      content: [{ type: 'text', text: `Screenshot saved to ${screenshotPath}` }],
    };
  }

• index-browser-only.js:130-147 (schema)

  Input schema definition for the screenshot tool, specifying parameters 'name' and 'fullPage'.

    name: 'screenshot',
    description: 'Take a screenshot of the current page',
    inputSchema: {
      type: 'object',
      properties: {
        name: {
          type: 'string',
          description: 'Name for the screenshot file',
          default: 'screenshot.png',
        },
        fullPage: {
          type: 'boolean',
          description: 'Capture full page',
          default: false,
        },
      },
    },
  },

• index-browser-only.js:182-183 (registration)

  Registration/dispatch in the CallToolRequestSchema handler: maps 'screenshot' tool calls to the screenshot handler method.

  case 'screenshot':
    return await this.screenshot(args.name, args.fullPage);

• arm64_browser_tools.py:61-105 (helper)

  Python helper wrapper that invokes the MCP 'screenshot' tool via subprocess on the JS server.

  def arm64_browser_screenshot(name: str = "screenshot.png", full_page: bool = False) -> str:
      """Take a screenshot using ARM64 Chromium browser.
      
      Args:
          name: Filename for the screenshot
          full_page: Whether to capture the full page
          
      Returns:
          Path to screenshot file or error message
      """
      request = {
          "jsonrpc": "2.0",
          "method": "tools/call",
          "params": {"name": "screenshot", "arguments": {"name": name, "fullPage": full_page}},
          "id": 1
      }
      
      try:
          result = subprocess.run(
              ["node", INDEX_PATH],
              input=json.dumps(request),
              text=True,
              capture_output=True,
              timeout=30,
              cwd=SERVER_DIR
          )
          
          # Parse both stdout and stderr for JSON responses
          all_output = result.stdout + result.stderr
          lines = [line for line in all_output.strip().split('\n') if line.startswith('{"')]
          
          for line in lines:
              try:
                  response = json.loads(line)
                  if 'result' in response:
                      content = response.get('result', {}).get('content', [{}])
                      return content[0].get('text', f'Screenshot saved as {name}')
              except json.JSONDecodeError:
                  continue
          
          return f"No valid response found. Output: {all_output[:200]}"
          
      except Exception as e:
          return f"Screenshot error: {e}"