@ai-sdk/anthropic 3.0.24 → 3.0.25

package/docs/05-anthropic.mdx

@@ -510,12 +510,13 @@ Parameters:
 The Computer Tool enables control of keyboard and mouse actions on a computer:
 
 ```ts
-const computerTool = anthropic.tools.computer_20241022({
+const computerTool = anthropic.tools.computer_20251124({
   displayWidthPx: 1920,
   displayHeightPx: 1080,
   displayNumber: 0, // Optional, for X11 environments
+  enableZoom: true, // Optional, enables the zoom action
 
-  execute: async ({ action, coordinate, text }) => {
+  execute: async ({ action, coordinate, text, region }) => {
     // Implement your computer control logic here
     // Return the result of the action
 
@@ -530,6 +531,13 @@ const computerTool = anthropic.tools.computer_20241022({
             .toString('base64'),
         };
       }
+      case 'zoom': {
+        // region is [x1, y1, x2, y2] defining the area to zoom into
+        return {
+          type: 'image',
+          data: fs.readFileSync('./data/zoomed-region.png').toString('base64'),
+        };
+      }
       default: {
         console.log('Action:', action);
         console.log('Coordinate:', coordinate);
@@ -548,13 +556,22 @@ const computerTool = anthropic.tools.computer_20241022({
 });
 ```
 
+<Note>
+  Use `computer_20251124` for Claude Opus 4.5 which supports the zoom action.
+  Use `computer_20250124` for Claude Sonnet 4.5, Haiku 4.5, Opus 4.1, Sonnet 4,
+  Opus 4, and Sonnet 3.7.
+</Note>
+
 Parameters:
 
-- `action` ('key' | 'type' | 'mouse_move' | 'left_click' | 'left_click_drag' | 'right_click' | 'middle_click' | 'double_click' | 'screenshot' | 'cursor_position'): The action to perform.
+- `action` ('key' | 'type' | 'mouse_move' | 'left_click' | 'left_click_drag' | 'right_click' | 'middle_click' | 'double_click' | 'screenshot' | 'cursor_position' | 'zoom'): The action to perform. The `zoom` action is only available with `computer_20251124`.
 - `coordinate` (number[], optional): Required for `mouse_move` and `left_click_drag` actions. Specifies the (x, y) coordinates.
 - `text` (string, optional): Required for `type` and `key` actions.
-
-These tools can be used in conjunction with the `sonnet-3-5-sonnet-20240620` model to enable more complex interactions and tasks.
+- `region` (number[], optional): Required for `zoom` action. Specifies `[x1, y1, x2, y2]` coordinates for the area to inspect.
+- `displayWidthPx` (number): The width of the display in pixels.
+- `displayHeightPx` (number): The height of the display in pixels.
+- `displayNumber` (number, optional): The display number for X11 environments.
+- `enableZoom` (boolean, optional): Enable the zoom action. Only available with `computer_20251124`. Default: `false`.
 
 ### Web Search Tool
 

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@ai-sdk/anthropic",
-  "version": "3.0.24",
+  "version": "3.0.25",
   "license": "Apache-2.0",
   "sideEffects": false,
   "main": "./dist/index.js",

package/src/anthropic-messages-api.ts

@@ -360,6 +360,15 @@ export type AnthropicTool =
       display_number: number;
       cache_control: AnthropicCacheControl | undefined;
     }
+  | {
+      name: string;
+      type: 'computer_20251124';
+      display_width_px: number;
+      display_height_px: number;
+      display_number: number;
+      enable_zoom?: boolean;
+      cache_control: AnthropicCacheControl | undefined;
+    }
   | {
       name: string;
       type:

package/src/anthropic-prepare-tools.ts

@@ -131,6 +131,19 @@ export async function prepareTools({
             });
             break;
           }
+          case 'anthropic.computer_20251124': {
+            betas.add('computer-use-2025-11-24');
+            anthropicTools.push({
+              name: 'computer',
+              type: 'computer_20251124',
+              display_width_px: tool.args.displayWidthPx as number,
+              display_height_px: tool.args.displayHeightPx as number,
+              display_number: tool.args.displayNumber as number,
+              enable_zoom: tool.args.enableZoom as boolean,
+              cache_control: undefined,
+            });
+            break;
+          }
           case 'anthropic.computer_20241022': {
             betas.add('computer-use-2024-10-22');
             anthropicTools.push({

package/src/anthropic-tools.ts

@@ -4,6 +4,7 @@ import { codeExecution_20250522 } from './tool/code-execution_20250522';
 import { codeExecution_20250825 } from './tool/code-execution_20250825';
 import { computer_20241022 } from './tool/computer_20241022';
 import { computer_20250124 } from './tool/computer_20250124';
+import { computer_20251124 } from './tool/computer_20251124';
 import { memory_20250818 } from './tool/memory_20250818';
 import { textEditor_20241022 } from './tool/text-editor_20241022';
 import { textEditor_20250124 } from './tool/text-editor_20250124';
@@ -77,6 +78,23 @@ export const anthropicTools = {
    */
   computer_20250124,
 
+  /**
+   * Claude can interact with computer environments through the computer use tool, which
+   * provides screenshot capabilities and mouse/keyboard control for autonomous desktop interaction.
+   *
+   * This version adds the zoom action for detailed screen region inspection.
+   *
+   * Image results are supported.
+   *
+   * Supported models: Claude Opus 4.5
+   *
+   * @param displayWidthPx - The width of the display being controlled by the model in pixels.
+   * @param displayHeightPx - The height of the display being controlled by the model in pixels.
+   * @param displayNumber - The display number to control (only relevant for X11 environments). If specified, the tool will be provided a display number in the tool definition.
+   * @param enableZoom - Enable zoom action. Set to true to allow Claude to zoom into specific screen regions. Default: false.
+   */
+  computer_20251124,
+
   /**
    * The memory tool enables Claude to store and retrieve information across conversations through a memory file directory.
    * Claude can create, read, update, and delete files that persist between sessions,

package/src/tool/computer_20251124.ts

@@ -0,0 +1,151 @@
+import {
+  createProviderToolFactory,
+  lazySchema,
+  zodSchema,
+} from '@ai-sdk/provider-utils';
+import { z } from 'zod/v4';
+
+const computer_20251124InputSchema = lazySchema(() =>
+  zodSchema(
+    z.object({
+      action: z.enum([
+        'key',
+        'hold_key',
+        'type',
+        'cursor_position',
+        'mouse_move',
+        'left_mouse_down',
+        'left_mouse_up',
+        'left_click',
+        'left_click_drag',
+        'right_click',
+        'middle_click',
+        'double_click',
+        'triple_click',
+        'scroll',
+        'wait',
+        'screenshot',
+        'zoom',
+      ]),
+      coordinate: z.tuple([z.number().int(), z.number().int()]).optional(),
+      duration: z.number().optional(),
+      region: z
+        .tuple([
+          z.number().int(),
+          z.number().int(),
+          z.number().int(),
+          z.number().int(),
+        ])
+        .optional(),
+      scroll_amount: z.number().optional(),
+      scroll_direction: z.enum(['up', 'down', 'left', 'right']).optional(),
+      start_coordinate: z
+        .tuple([z.number().int(), z.number().int()])
+        .optional(),
+      text: z.string().optional(),
+    }),
+  ),
+);
+
+export const computer_20251124 = createProviderToolFactory<
+  {
+    /**
+     * - `key`: Press a key or key-combination on the keyboard.
+     *   - This supports xdotool's `key` syntax.
+     *   - Examples: "a", "Return", "alt+Tab", "ctrl+s", "Up", "KP_0" (for the numpad 0 key).
+     * - `hold_key`: Hold down a key or multiple keys for a specified duration (in seconds). Supports the same syntax as `key`.
+     * - `type`: Type a string of text on the keyboard.
+     * - `cursor_position`: Get the current (x, y) pixel coordinate of the cursor on the screen.
+     * - `mouse_move`: Move the cursor to a specified (x, y) pixel coordinate on the screen.
+     * - `left_mouse_down`: Press the left mouse button.
+     * - `left_mouse_up`: Release the left mouse button.
+     * - `left_click`: Click the left mouse button at the specified (x, y) pixel coordinate on the screen. You can also include a key combination to hold down while clicking using the `text` parameter.
+     * - `left_click_drag`: Click and drag the cursor from `start_coordinate` to a specified (x, y) pixel coordinate on the screen.
+     * - `right_click`: Click the right mouse button at the specified (x, y) pixel coordinate on the screen.
+     * - `middle_click`: Click the middle mouse button at the specified (x, y) pixel coordinate on the screen.
+     * - `double_click`: Double-click the left mouse button at the specified (x, y) pixel coordinate on the screen.
+     * - `triple_click`: Triple-click the left mouse button at the specified (x, y) pixel coordinate on the screen.
+     * - `scroll`: Scroll the screen in a specified direction by a specified amount of clicks of the scroll wheel, at the specified (x, y) pixel coordinate. DO NOT use PageUp/PageDown to scroll.
+     * - `wait`: Wait for a specified duration (in seconds).
+     * - `screenshot`: Take a screenshot of the screen.
+     * - `zoom`: View a specific region of the screen at full resolution. Requires `enableZoom: true` in tool definition. Takes a `region` parameter with coordinates `[x1, y1, x2, y2]` defining top-left and bottom-right corners of the area to inspect.
+     */
+    action:
+      | 'key'
+      | 'hold_key'
+      | 'type'
+      | 'cursor_position'
+      | 'mouse_move'
+      | 'left_mouse_down'
+      | 'left_mouse_up'
+      | 'left_click'
+      | 'left_click_drag'
+      | 'right_click'
+      | 'middle_click'
+      | 'double_click'
+      | 'triple_click'
+      | 'scroll'
+      | 'wait'
+      | 'screenshot'
+      | 'zoom';
+
+    /**
+     * (x, y): The x (pixels from the left edge) and y (pixels from the top edge) coordinates to move the mouse to. Required only by `action=mouse_move` and `action=left_click_drag`.
+     */
+    coordinate?: [number, number];
+
+    /**
+     * The duration to hold the key down for. Required only by `action=hold_key` and `action=wait`.
+     */
+    duration?: number;
+
+    /**
+     * [x1, y1, x2, y2]: The coordinates defining the region to zoom into. x1, y1 is the top-left corner and x2, y2 is the bottom-right corner. Required only by `action=zoom`.
+     */
+    region?: [number, number, number, number];
+
+    /**
+     * The number of 'clicks' to scroll. Required only by `action=scroll`.
+     */
+    scroll_amount?: number;
+
+    /**
+     * The direction to scroll the screen. Required only by `action=scroll`.
+     */
+    scroll_direction?: 'up' | 'down' | 'left' | 'right';
+
+    /**
+     * (x, y): The x (pixels from the left edge) and y (pixels from the top edge) coordinates to start the drag from. Required only by `action=left_click_drag`.
+     */
+    start_coordinate?: [number, number];
+
+    /**
+     * Required only by `action=type`, `action=key`, and `action=hold_key`. Can also be used by click or scroll actions to hold down keys while clicking or scrolling.
+     */
+    text?: string;
+  },
+  {
+    /**
+     * The width of the display being controlled by the model in pixels.
+     */
+    displayWidthPx: number;
+
+    /**
+     * The height of the display being controlled by the model in pixels.
+     */
+    displayHeightPx: number;
+
+    /**
+     * The display number to control (only relevant for X11 environments). If specified, the tool will be provided a display number in the tool definition.
+     */
+    displayNumber?: number;
+
+    /**
+     * Enable zoom action. Set to true to allow Claude to zoom into specific screen regions. Default: false.
+     */
+    enableZoom?: boolean;
+  }
+>({
+  id: 'anthropic.computer_20251124',
+  inputSchema: computer_20251124InputSchema,
+});