chore: introduce capabilities argument (#135)

src/tools/common.ts

@@ -24,6 +24,7 @@ const waitSchema = z.object({
 });
 
 const wait: Tool = {
+  capability: 'wait',
   schema: {
     name: 'browser_wait',
     description: 'Wait for a specified time in seconds',
@@ -44,6 +45,7 @@ const wait: Tool = {
 const closeSchema = z.object({});
 
 const close: Tool = {
+  capability: 'core',
   schema: {
     name: 'browser_close',
     description: 'Close the page',

src/tools/files.ts

@@ -19,18 +19,19 @@ import { zodToJsonSchema } from 'zod-to-json-schema';
 
 import type { ToolFactory } from './tool';
 
-const chooseFileSchema = z.object({
+const uploadFileSchema = z.object({
   paths: z.array(z.string()).describe('The absolute paths to the files to upload. Can be a single file or multiple files.'),
 });
 
-const chooseFile: ToolFactory = captureSnapshot => ({
+const uploadFile: ToolFactory = captureSnapshot => ({
+  capability: 'files',
   schema: {
-    name: 'browser_choose_file',
-    description: 'Choose one or multiple files to upload',
-    inputSchema: zodToJsonSchema(chooseFileSchema),
+    name: 'browser_file_upload',
+    description: 'Upload one or multiple files',
+    inputSchema: zodToJsonSchema(uploadFileSchema),
   },
   handle: async (context, params) => {
-    const validatedParams = chooseFileSchema.parse(params);
+    const validatedParams = uploadFileSchema.parse(params);
     const tab = context.currentTab();
     return await tab.runAndWait(async () => {
       await tab.submitFileChooser(validatedParams.paths);
@@ -43,5 +44,5 @@ const chooseFile: ToolFactory = captureSnapshot => ({
 });
 
 export default (captureSnapshot: boolean) => [
-  chooseFile(captureSnapshot),
+  uploadFile(captureSnapshot),
 ];

src/tools/install.ts

@@ -23,6 +23,7 @@ import { zodToJsonSchema } from 'zod-to-json-schema';
 import type { Tool } from './tool';
 
 const install: Tool = {
+  capability: 'install',
   schema: {
     name: 'browser_install',
     description: 'Install the browser specified in the config. Call this if you get an error about the browser not being installed.',

src/tools/keyboard.ts

@@ -24,6 +24,7 @@ const pressKeySchema = z.object({
 });
 
 const pressKey: ToolFactory = captureSnapshot => ({
+  capability: 'core',
   schema: {
     name: 'browser_press_key',
     description: 'Press a key on the keyboard',

src/tools/navigate.ts

@@ -24,6 +24,7 @@ const navigateSchema = z.object({
 });
 
 const navigate: ToolFactory = captureSnapshot => ({
+  capability: 'core',
   schema: {
     name: 'browser_navigate',
     description: 'Navigate to a URL',
@@ -44,6 +45,7 @@ const navigate: ToolFactory = captureSnapshot => ({
 const goBackSchema = z.object({});
 
 const goBack: ToolFactory = snapshot => ({
+  capability: 'history',
   schema: {
     name: 'browser_navigate_back',
     description: 'Go back to the previous page',
@@ -62,6 +64,7 @@ const goBack: ToolFactory = snapshot => ({
 const goForwardSchema = z.object({});
 
 const goForward: ToolFactory = snapshot => ({
+  capability: 'history',
   schema: {
     name: 'browser_navigate_forward',
     description: 'Go forward to the next page',

src/tools/pdf.ts

@@ -27,6 +27,7 @@ import type { Tool } from './tool';
 const pdfSchema = z.object({});
 
 const pdf: Tool = {
+  capability: 'pdf',
   schema: {
     name: 'browser_pdf_save',
     description: 'Save page as PDF',

src/tools/screen.ts

@@ -20,6 +20,7 @@ import { zodToJsonSchema } from 'zod-to-json-schema';
 import type { Tool } from './tool';
 
 const screenshot: Tool = {
+  capability: 'core',
   schema: {
     name: 'browser_screen_capture',
     description: 'Take a screenshot of the current page',
@@ -45,6 +46,7 @@ const moveMouseSchema = elementSchema.extend({
 });
 
 const moveMouse: Tool = {
+  capability: 'core',
   schema: {
     name: 'browser_screen_move_mouse',
     description: 'Move mouse to a given position',
@@ -67,6 +69,7 @@ const clickSchema = elementSchema.extend({
 });
 
 const click: Tool = {
+  capability: 'core',
   schema: {
     name: 'browser_screen_click',
     description: 'Click left mouse button',
@@ -93,6 +96,7 @@ const dragSchema = elementSchema.extend({
 });
 
 const drag: Tool = {
+  capability: 'core',
   schema: {
     name: 'browser_screen_drag',
     description: 'Drag left mouse button',
@@ -118,6 +122,7 @@ const typeSchema = z.object({
 });
 
 const type: Tool = {
+  capability: 'core',
   schema: {
     name: 'browser_screen_type',
     description: 'Type text',

src/tools/snapshot.ts

@@ -21,6 +21,7 @@ import type * as playwright from 'playwright';
 import type { Tool } from './tool';
 
 const snapshot: Tool = {
+  capability: 'core',
   schema: {
     name: 'browser_snapshot',
     description: 'Capture accessibility snapshot of the current page, this is better than screenshot',
@@ -38,6 +39,7 @@ const elementSchema = z.object({
 });
 
 const click: Tool = {
+  capability: 'core',
   schema: {
     name: 'browser_click',
     description: 'Perform click on a web page',
@@ -63,6 +65,7 @@ const dragSchema = z.object({
 });
 
 const drag: Tool = {
+  capability: 'core',
   schema: {
     name: 'browser_drag',
     description: 'Perform drag and drop between two elements',
@@ -82,6 +85,7 @@ const drag: Tool = {
 };
 
 const hover: Tool = {
+  capability: 'core',
   schema: {
     name: 'browser_hover',
     description: 'Hover over element on page',
@@ -106,6 +110,7 @@ const typeSchema = elementSchema.extend({
 });
 
 const type: Tool = {
+  capability: 'core',
   schema: {
     name: 'browser_type',
     description: 'Type text into editable element',
@@ -133,6 +138,7 @@ const selectOptionSchema = elementSchema.extend({
 });
 
 const selectOption: Tool = {
+  capability: 'core',
   schema: {
     name: 'browser_select_option',
     description: 'Select an option in a dropdown',
@@ -155,6 +161,7 @@ const screenshotSchema = z.object({
 });
 
 const screenshot: Tool = {
+  capability: 'core',
   schema: {
     name: 'browser_take_screenshot',
     description: `Take a screenshot of the current page. You can't perform actions based on the screenshot, use browser_snapshot for actions.`,

src/tools/tabs.ts

@@ -20,6 +20,7 @@ import { zodToJsonSchema } from 'zod-to-json-schema';
 import type { ToolFactory, Tool } from './tool';
 
 const listTabs: Tool = {
+  capability: 'tabs',
   schema: {
     name: 'browser_tab_list',
     description: 'List browser tabs',
@@ -40,6 +41,7 @@ const selectTabSchema = z.object({
 });
 
 const selectTab: ToolFactory = captureSnapshot => ({
+  capability: 'tabs',
   schema: {
     name: 'browser_tab_select',
     description: 'Select a tab by index',
@@ -58,6 +60,7 @@ const newTabSchema = z.object({
 });
 
 const newTab: Tool = {
+  capability: 'tabs',
   schema: {
     name: 'browser_tab_new',
     description: 'Open a new tab',
@@ -77,6 +80,7 @@ const closeTabSchema = z.object({
 });
 
 const closeTab: ToolFactory = captureSnapshot => ({
+  capability: 'tabs',
   schema: {
     name: 'browser_tab_close',
     description: 'Close a tab',

src/tools/tool.ts

@@ -18,6 +18,8 @@ import type { ImageContent, TextContent } from '@modelcontextprotocol/sdk/types'
 import type { JsonSchema7Type } from 'zod-to-json-schema';
 import type { Context } from '../context';
 
+export type ToolCapability = 'core' | 'tabs' | 'pdf' | 'history' | 'wait' | 'files' | 'install';
+
 export type ToolSchema = {
   name: string;
   description: string;
@@ -30,6 +32,7 @@ export type ToolResult = {
 };
 
 export type Tool = {
+  capability: ToolCapability;
   schema: ToolSchema;
   handle: (context: Context, params?: Record<string, any>) => Promise<ToolResult>;
 };