chore: tune press/type via cli (#38898)

Author: pavelfeldman

packages/playwright/src/mcp/browser/tools/keyboard.ts

@@ -32,13 +32,37 @@ const pressKey = defineTabTool({
   },
 
   handle: async (tab, params, response) => {
-    response.setIncludeSnapshot();
     response.addCode(`// Press ${params.key}`);
     response.addCode(`await page.keyboard.press('${params.key}');`);
+    await tab.page.keyboard.press(params.key);
+  },
+});
 
-    await tab.waitForCompletion(async () => {
-      await tab.page.keyboard.press(params.key);
-    });
+const pressSequentially = defineTabTool({
+  capability: 'internal',
+
+  schema: {
+    name: 'browser_press_sequentially',
+    title: 'Press sequentially',
+    description: 'Press text sequentially on the keyboard',
+    inputSchema: z.object({
+      text: z.string().describe('Text to press sequentially'),
+      submit: z.boolean().optional().describe('Whether to submit entered text (press Enter after)'),
+    }),
+    type: 'input',
+  },
+
+  handle: async (tab, params, response) => {
+    response.addCode(`// Press ${params.text}`);
+    response.addCode(`await page.keyboard.type('${params.text}');`);
+    await tab.page.keyboard.type(params.text);
+    if (params.submit) {
+      response.addCode(`await page.keyboard.press('Enter');`);
+      response.setIncludeSnapshot();
+      await tab.waitForCompletion(async () => {
+        await tab.page.keyboard.press('Enter');
+      });
+    }
   },
 });
 
@@ -84,4 +108,5 @@ const type = defineTabTool({
 export default [
   pressKey,
   type,
+  pressSequentially,
 ];

packages/playwright/src/mcp/terminal/commands.ts

@@ -156,7 +156,7 @@ const networkRequests = declareCommand({
 });
 
 const pressKey = declareCommand({
-  name: 'press-key',
+  name: 'press',
   description: 'Press a key on the keyboard',
   args: z.object({
     key: z.string().describe('Name of the key to press or a character to generate, such as `ArrowLeft` or `a`'),
@@ -226,15 +226,13 @@ const type = declareCommand({
   name: 'type',
   description: 'Type text into editable element',
   args: z.object({
-    ref: z.string().describe('Exact target element reference from the page snapshot'),
     text: z.string().describe('Text to type into the element'),
   }),
   options: z.object({
     submit: z.boolean().optional().describe('Whether to submit entered text (press Enter after)'),
-    slowly: z.boolean().optional().describe('Whether to type one character at a time. Useful for triggering key handlers in the page.'),
   }),
-  toolName: 'browser_type',
-  toolParams: ({ ref, text }, { submit, slowly }) => ({ ref, text, submit, slowly }),
+  toolName: 'browser_press_sequentially',
+  toolParams: ({ text }, { submit }) => ({ text, submit }),
 });
 
 const waitFor = declareCommand({

packages/playwright/src/mcp/terminal/help.json

@@ -1,5 +1,5 @@
 {
-  "global": "Usage: playwright-cli <command> [options]\nCommands:\n  click <ref>                             perform click on a web page\n  close                                   close the page\n  dblclick <ref>                          perform double click on a web page\n  console <level>                         returns all console messages\n  drag <startRef> <endRef>                perform drag and drop between two elements\n  evaluate <function> <ref>               evaluate javascript expression on page or element\n  upload-file                             upload one or multiple files\n  handle-dialog <accept> <promptText>     handle a dialog\n  hover <ref>                             hover over element on page\n  open <url>                              open url\n  go-back                                 go back to the previous page\n  network-requests                        returns all network requests since loading the page\n  press-key <key>                         press a key on the keyboard\n  resize <width> <height>                 resize the browser window\n  run-code <code>                         run playwright code snippet\n  select-option <ref> <values>            select an option in a dropdown\n  snapshot                                capture accessibility snapshot of the current page, this is better than screenshot\n  screenshot <ref>                        take a screenshot of the current page. you can't perform actions based on the screenshot, use browser_snapshot for actions.\n  type <ref> <text>                       type text into editable element\n  wait-for                                wait for text to appear or disappear or a specified time to pass\n  tab <action> <index>                    close a browser tab\n  mouse-click-xy <x> <y>                  click left mouse button at a given position\n  mouse-drag-xy <startX> <startY> <endX> <endY> drag left mouse button to a given position\n  mouse-move-xy <x> <y>                   move mouse to a given position\n  pdf-save                                save page as pdf\n  start-tracing                           start trace recording\n  stop-tracing                            stop trace recording",
+  "global": "Usage: playwright-cli <command> [options]\nCommands:\n  click <ref>                             perform click on a web page\n  close                                   close the page\n  dblclick <ref>                          perform double click on a web page\n  console <level>                         returns all console messages\n  drag <startRef> <endRef>                perform drag and drop between two elements\n  evaluate <function> <ref>               evaluate javascript expression on page or element\n  upload-file                             upload one or multiple files\n  handle-dialog <accept> <promptText>     handle a dialog\n  hover <ref>                             hover over element on page\n  open <url>                              open url\n  go-back                                 go back to the previous page\n  network-requests                        returns all network requests since loading the page\n  press <key>                             press a key on the keyboard\n  resize <width> <height>                 resize the browser window\n  run-code <code>                         run playwright code snippet\n  select-option <ref> <values>            select an option in a dropdown\n  snapshot                                capture accessibility snapshot of the current page, this is better than screenshot\n  screenshot <ref>                        take a screenshot of the current page. you can't perform actions based on the screenshot, use browser_snapshot for actions.\n  type <text>                             type text into editable element\n  wait-for                                wait for text to appear or disappear or a specified time to pass\n  tab <action> <index>                    close a browser tab\n  mouse-click-xy <x> <y>                  click left mouse button at a given position\n  mouse-drag-xy <startX> <startY> <endX> <endY> drag left mouse button to a given position\n  mouse-move-xy <x> <y>                   move mouse to a given position\n  pdf-save                                save page as pdf\n  start-tracing                           start trace recording\n  stop-tracing                            stop trace recording",
   "commands": {
     "click": "playwright-cli click <ref>\n\nPerform click on a web page\n\nArguments:\n  <ref>\tExact target element reference from the page snapshot\nOptions:\n  --button\tbutton to click, defaults to left\n  --modifiers\tmodifier keys to press",
     "close": "playwright-cli close \n\nClose the page\n",
@@ -13,13 +13,13 @@
     "open": "playwright-cli open <url>\n\nOpen URL\n\nArguments:\n  <url>\tThe URL to navigate to\nOptions:\n  --headed\trun browser in headed mode",
     "go-back": "playwright-cli go-back \n\nGo back to the previous page\n",
     "network-requests": "playwright-cli network-requests \n\nReturns all network requests since loading the page\n\nOptions:\n  --includeStatic\twhether to include successful static resources like images, fonts, scripts, etc. defaults to false.",
-    "press-key": "playwright-cli press-key <key>\n\nPress a key on the keyboard\n\nArguments:\n  <key>\tName of the key to press or a character to generate, such as `ArrowLeft` or `a`",
+    "press": "playwright-cli press <key>\n\nPress a key on the keyboard\n\nArguments:\n  <key>\tName of the key to press or a character to generate, such as `ArrowLeft` or `a`",
     "resize": "playwright-cli resize <width> <height>\n\nResize the browser window\n\nArguments:\n  <width>\tWidth of the browser window\n  <height>\tHeight of the browser window",
     "run-code": "playwright-cli run-code <code>\n\nRun Playwright code snippet\n\nArguments:\n  <code>\tA JavaScript function containing Playwright code to execute. It will be invoked with a single argument, page, which you can use for any page interaction.",
     "select-option": "playwright-cli select-option <ref> <values>\n\nSelect an option in a dropdown\n\nArguments:\n  <ref>\tExact target element reference from the page snapshot\n  <values>\tArray of values to select in the dropdown. This can be a single value or multiple values.",
     "snapshot": "playwright-cli snapshot \n\nCapture accessibility snapshot of the current page, this is better than screenshot\n\nOptions:\n  --filename\tsave snapshot to markdown file instead of returning it in the response.",
     "screenshot": "playwright-cli screenshot <ref>\n\nTake a screenshot of the current page. You can't perform actions based on the screenshot, use browser_snapshot for actions.\n\nArguments:\n  <ref>\tExact target element reference from the page snapshot.\nOptions:\n  --filename\tfile name to save the screenshot to. defaults to `page-{timestamp}.{png|jpeg}` if not specified.\n  --fullPage\twhen true, takes a screenshot of the full scrollable page, instead of the currently visible viewport.",
-    "type": "playwright-cli type <ref> <text>\n\nType text into editable element\n\nArguments:\n  <ref>\tExact target element reference from the page snapshot\n  <text>\tText to type into the element\nOptions:\n  --submit\twhether to submit entered text (press enter after)\n  --slowly\twhether to type one character at a time. useful for triggering key handlers in the page.",
+    "type": "playwright-cli type <text>\n\nType text into editable element\n\nArguments:\n  <text>\tText to type into the element\nOptions:\n  --submit\twhether to submit entered text (press enter after)",
     "wait-for": "playwright-cli wait-for \n\nWait for text to appear or disappear or a specified time to pass\n\nOptions:\n  --time\tthe time to wait in seconds\n  --text\tthe text to wait for\n  --textGone\tthe text to wait for to disappear",
     "tab": "playwright-cli tab <action> <index>\n\nClose a browser tab\n\nArguments:\n  <action>\tAction to perform on tabs, 'list' | 'new' | 'close' | 'select'\n  <index>\tTab index. If omitted, current tab is closed.",
     "mouse-click-xy": "playwright-cli mouse-click-xy <x> <y>\n\nClick left mouse button at a given position\n\nArguments:\n  <x>\tX coordinate\n  <y>\tY coordinate",

tests/mcp/generator.spec.ts

@@ -47,6 +47,7 @@ test('generator tools intent', async ({ startClient }) => {
     'browser_navigate_back',
     'browser_open',
     'browser_press_key',
+    'browser_press_sequentially',
     'browser_resize',
     'browser_run_code',
     'browser_select_option',