test-wuying-agentbay-sdk 0.13.0-beta.20251212114304 → 0.13.0-beta.20251212140302

package/docs/api/browser-use/browser-agent.md

@@ -7,16 +7,19 @@
 
 - [act](#act)
 - [actAsync](#actasync)
+- [close](#close)
 - [extract](#extract)
 - [extractAsync](#extractasync)
+- [navigate](#navigate)
 - [observe](#observe)
 - [observeAsync](#observeasync)
+- [screenshot](#screenshot)
 
 ## Methods
 
 ### act
 
-▸ **act**(`options`, `page`): `Promise`\<``ActResult``\>
+▸ **act**(`options`, `page?`): `Promise`\<``ActResult``\>
 
 ------------------ ACT ------------------ *
 
@@ -25,7 +28,7 @@
 | Name | Type |
 | :------ | :------ |
 | `options` | ``ActOptions`` |
-| `page` | `any` |
+| `page?` | `any` |
 
 #### Returns
 
@@ -35,14 +38,14 @@ ___
 
 ### actAsync
 
-▸ **actAsync**(`options`, `page`): `Promise`\<``ActResult``\>
+▸ **actAsync**(`options`, `page?`): `Promise`\<``ActResult``\>
 
 #### Parameters
 
 | Name | Type |
 | :------ | :------ |
 | `options` | ``ActOptions`` |
-| `page` | `any` |
+| `page?` | `any` |
 
 #### Returns
 
@@ -50,57 +53,87 @@ ___
 
 ___
 
+### close
+
+▸ **close**(): `Promise`\<`boolean`\>
+
+------------------ CLOSE ------------------ *
+
+#### Returns
+
+`Promise`\<`boolean`\>
+
+___
+
 ### extract
 
-▸ **extract**\<`T`\>(`options`, `page`): `Promise`\<[`boolean`, ``null`` \| `T`]\>
+▸ **extract**\<`TSchema`\>(`options`, `page?`): `Promise`\<[`boolean`, ``null`` \| `TypeOf`\<`TSchema`\>]\>
 
 ------------------ EXTRACT ------------------ *
 
 #### Type parameters
 
-| Name |
-| :------ |
-| `T` |
+| Name | Type |
+| :------ | :------ |
+| `TSchema` | extends `ZodType`\<`any`, `any`, `any`, `TSchema`\> |
 
 #### Parameters
 
 | Name | Type |
 | :------ | :------ |
-| `options` | ``ExtractOptions``\<`T`\> |
-| `page` | `any` |
+| `options` | ``ExtractOptions``\<`TSchema`\> |
+| `page?` | `any` |
 
 #### Returns
 
-`Promise`\<[`boolean`, ``null`` \| `T`]\>
+`Promise`\<[`boolean`, ``null`` \| `TypeOf`\<`TSchema`\>]\>
 
 ___
 
 ### extractAsync
 
-▸ **extractAsync**\<`T`\>(`options`, `page`): `Promise`\<[`boolean`, ``null`` \| `T`]\>
+▸ **extractAsync**\<`TSchema`\>(`options`, `page?`): `Promise`\<[`boolean`, ``null`` \| `TypeOf`\<`TSchema`\>]\>
 
 #### Type parameters
 
-| Name |
-| :------ |
-| `T` |
+| Name | Type |
+| :------ | :------ |
+| `TSchema` | extends `ZodType`\<`any`, `any`, `any`, `TSchema`\> |
+
+#### Parameters
+
+| Name | Type |
+| :------ | :------ |
+| `options` | ``ExtractOptions``\<`TSchema`\> |
+| `page?` | `any` |
+
+#### Returns
+
+`Promise`\<[`boolean`, ``null`` \| `TypeOf`\<`TSchema`\>]\>
+
+___
+
+### navigate
+
+▸ **navigate**(`url`): `Promise`\<`string`\>
+
+------------------ NAVIGATE ------------------ *
 
 #### Parameters
 
 | Name | Type |
 | :------ | :------ |
-| `options` | ``ExtractOptions``\<`T`\> |
-| `page` | `any` |
+| `url` | `string` |
 
 #### Returns
 
-`Promise`\<[`boolean`, ``null`` \| `T`]\>
+`Promise`\<`string`\>
 
 ___
 
 ### observe
 
-▸ **observe**(`options`, `page`): `Promise`\<``boolean`, [`ObserveResult``[]]\>
+▸ **observe**(`options`, `page?`): `Promise`\<``boolean`, [`ObserveResult``[]]\>
 
 ------------------ OBSERVE ------------------ *
 
@@ -109,7 +142,7 @@ ___
 | Name | Type |
 | :------ | :------ |
 | `options` | ``ObserveOptions`` |
-| `page` | `any` |
+| `page?` | `any` |
 
 #### Returns
 
@@ -119,15 +152,37 @@ ___
 
 ### observeAsync
 
-▸ **observeAsync**(`options`, `page`): `Promise`\<``boolean`, [`ObserveResult``[]]\>
+▸ **observeAsync**(`options`, `page?`): `Promise`\<``boolean`, [`ObserveResult``[]]\>
 
 #### Parameters
 
 | Name | Type |
 | :------ | :------ |
 | `options` | ``ObserveOptions`` |
-| `page` | `any` |
+| `page?` | `any` |
 
 #### Returns
 
 `Promise`\<``boolean`, [`ObserveResult``[]]\>
+
+___
+
+### screenshot
+
+▸ **screenshot**(`page?`, `full_page?`, `quality?`, `clip?`, `timeout?`): `Promise`\<`string`\>
+
+------------------ SCREENSHOT ------------------ *
+
+#### Parameters
+
+| Name | Type | Default value |
+| :------ | :------ | :------ |
+| `page` | `any` | `null` |
+| `full_page` | `boolean` | `true` |
+| `quality` | `number` | `80` |
+| `clip?` | `Record`\<`string`, `number`\> | `undefined` |
+| `timeout?` | `number` | `undefined` |
+
+#### Returns
+
+`Promise`\<`string`\>

package/docs/examples/browser-use/browser/basic-usage.ts

@@ -3,14 +3,19 @@
  * This example demonstrates the core browser operations without external dependencies.
  */
 
-import { AgentBay, CreateSessionParams, ActOptions, ObserveOptions, ExtractOptions } from 'wuying-agentbay-sdk';
-
-// Simple schema for demonstration
-class PageInfo {
-  title: string = "";
-  url: string = "";
-}
-
+import {
+  AgentBay,
+  CreateSessionParams,
+  ActOptions,
+  ObserveOptions,
+  ExtractOptions,
+} from "wuying-agentbay-sdk";
+import { z } from "zod";
+
+const PageInfoSchema = z.object({
+  title: z.string().describe("page title"),
+  url: z.string().describe("page url"),
+});
 async function main() {
   // Get API key from environment variable
   const apiKey = process.env.AGENTBAY_API_KEY;
@@ -57,7 +62,7 @@ async function main() {
     // Example: Mock page object (in real usage, you'd get this from Playwright)
     const mockPage = {
       url: () => "https://example.com",
-      title: () => "Example Domain"
+      title: () => "Example Domain",
     };
 
     // Example 1: Perform an action
@@ -65,7 +70,7 @@ async function main() {
     try {
       const actOptions: ActOptions = {
         action: "Click the 'More information...' link",
-        timeoutMS: 5000
+        timeout: 5,
       };
 
       const actResult = await session.browser.agent.act(actOptions, mockPage);
@@ -81,10 +86,11 @@ async function main() {
     console.log("\n--- Example 2: Observing page elements ---");
     try {
       const observeOptions: ObserveOptions = {
-        instruction: "Find all links and buttons on the page"
+        instruction: "Find all links and buttons on the page",
       };
 
-      const [observeSuccess, observations] = await session.browser.agent.observe(observeOptions, mockPage);
+      const [observeSuccess, observations] =
+        await session.browser.agent.observe(observeOptions, mockPage);
       console.log("Observe success:", observeSuccess);
       console.log("Number of observations:", observations.length);
 
@@ -92,7 +98,7 @@ async function main() {
         console.log(`Observation ${index + 1}:`, {
           selector: obs.selector,
           description: obs.description,
-          method: obs.method
+          method: obs.method,
         });
       });
     } catch (error) {
@@ -102,22 +108,24 @@ async function main() {
     // Example 3: Extract structured data from the page
     console.log("\n--- Example 3: Extracting structured data ---");
     try {
-      const extractOptions: ExtractOptions<PageInfo> = {
+      const extractOptions: ExtractOptions<typeof PageInfoSchema> = {
         instruction: "Extract the page title and URL",
-        schema: PageInfo,
-        use_text_extract: false
+        schema: PageInfoSchema,
+        use_text_extract: false,
       };
 
-      const [extractSuccess, extractedData] = await session.browser.agent.extract(extractOptions, mockPage);
+      const [extractSuccess, extracted] = await session.browser.agent.extract(
+        extractOptions,
+        mockPage
+      );
       console.log("Extract success:", extractSuccess);
-      console.log("Extracted data count:", extractedData.length);
 
-      extractedData.forEach((data: PageInfo, index: number) => {
-        console.log(`Extracted item ${index + 1}:`, {
-          title: data.title,
-          url: data.url
+      if (extractSuccess && extracted) {
+        console.log("Extracted data:", {
+          title: extracted.title,
+          url: extracted.url,
         });
-      });
+      }
     } catch (error) {
       console.log("Extraction failed:", error);
     }
@@ -125,7 +133,6 @@ async function main() {
     // Clean up
     console.log("\n--- Cleanup ---");
     console.log("Browser session completed successfully");
-
   } catch (error) {
     console.error("Error in main function:", error);
   }

package/docs/examples/browser-use/browser/run-2048.ts

@@ -6,35 +6,29 @@
  * - Utilize PageUseAgent to run 2048 game
  */
 
-import { AgentBay, CreateSessionParams } from 'wuying-agentbay-sdk';
-import { BrowserOption, ExtractOptions, ActOptions } from 'wuying-agentbay-sdk/dist/browser';
-// import { chromium } from 'playwright';
-
-class GameState {
-  score?: number;
-  highestTile?: number;
-  grid: number[][] = [];
-}
+import { AgentBay, CreateSessionParams } from "wuying-agentbay-sdk";
+import { BrowserOption, ExtractOptions } from "wuying-agentbay-sdk";
+import { z } from "zod";
 
-class MoveAnalysis {
-  move?: number;
-  confidence: number = 0;
-  reasoning: string = "";
-}
+const GameStateSchema = z.object({
+  score: z.number().optional(),
+  highestTile: z.number().optional(),
+  grid: z.array(z.array(z.number())),
+});
 
 function transposeGrid(grid: number[][]): number[][] {
   if (!grid || grid.length === 0) {
     return [];
   }
-  return grid[0].map((_, colIndex) => grid.map(row => row[colIndex]));
+  return grid[0].map((_, colIndex) => grid.map((row) => row[colIndex]));
 }
 
 function formatGridForLlmInstruction(gridData: number[][]): string {
   const formattedRows: string[] = [];
   for (let i = 0; i < gridData.length; i++) {
-    formattedRows.push(`row${i + 1}: [${gridData[i].join(', ')}]`);
+    formattedRows.push(`row${i + 1}: [${gridData[i].join(", ")}]`);
   }
-  return formattedRows.join('\n');
+  return formattedRows.join("\n");
 }
 
 async function main() {
@@ -66,17 +60,17 @@ async function main() {
       console.log("endpoint_url =", endpointUrl);
 
       // Note: Install playwright with: npm install playwright
-      const { chromium } = require('playwright');
+      const { chromium } = require("playwright");
       const browser = await chromium.connectOverCDP(endpointUrl);
       let page = null;
 
       try {
-        const context = browser.contexts()[0]
+        const context = browser.contexts()[0];
         page = await context.newPage();
         console.log("🌐 Navigating to 2048...");
         await page.goto("https://ovolve.github.io/2048-AI/", {
           waitUntil: "domcontentloaded",
-          timeout: 180000
+          timeout: 180000,
         });
         console.log("🌐 Navigated to 2048 done");
         await page.waitForSelector(".grid-container", { timeout: 10000 });
@@ -87,11 +81,11 @@ async function main() {
 
         while (true) {
           console.log("🔄 Game loop iteration...");
-          await new Promise(resolve => setTimeout(resolve, 300));
+          await new Promise((resolve) => setTimeout(resolve, 300));
 
           // Get current game state
           console.log("📊 Extracting game state...");
-          const gameStateOptions: ExtractOptions<GameState> = {
+          const gameStateOptions: ExtractOptions<typeof GameStateSchema> = {
             instruction: `
               Extract the current game state:
               1. Score from the score counter
@@ -105,20 +99,25 @@ async function main() {
                  For instance, if the only tiles present are the two above, the grid should be:[[0, 0, 0, 2], [0, 0, 0, 0], [0, 0, 0, 0], [2, 0, 0, 0]]
               3. Highest tile value present
             `,
-            schema: GameState,
-            use_text_extract: false
-
+            schema: GameStateSchema,
+            use_text_extract: false,
           };
 
-          const [success, gameStates] = await session.browser.agent.extract(gameStateOptions, page);
-          if (success && gameStates.length > 0) {
-            const gameState = gameStates[0];
+          const [success, gameState] = await session.browser.agent.extract(
+            gameStateOptions,
+            page
+          );
+          if (success && gameState) {
             const transposedGrid = transposeGrid(gameState.grid);
             console.log(`transposed grid: ${JSON.stringify(transposedGrid)}`);
             console.log(`gameState: ${JSON.stringify(gameState)}`);
             const gridInstruction = formatGridForLlmInstruction(transposedGrid);
 
-            if (lastTransposedGrid !== null && JSON.stringify(transposedGrid) === JSON.stringify(lastTransposedGrid)) {
+            if (
+              lastTransposedGrid !== null &&
+              JSON.stringify(transposedGrid) ===
+                JSON.stringify(lastTransposedGrid)
+            ) {
               transposedGridNotChangedTimes += 1;
             } else {
               transposedGridNotChangedTimes = 0;
@@ -144,22 +143,33 @@ async function main() {
 
             if (transposedGridNotChangedTimes >= 1) {
               instructionStr += `
-                9. Do not generate move value in ${JSON.stringify(lastMoveHistory)}
-                10. If last move value ${lastMoveHistory[lastMoveHistory.length - 1]} moves up or down, then generate move value with left or right direction, otherwise generate move value with up or down direction
+                9. Do not generate move value in ${JSON.stringify(
+                  lastMoveHistory
+                )}
+                10. If last move value ${
+                  lastMoveHistory[lastMoveHistory.length - 1]
+                } moves up or down, then generate move value with left or right direction, otherwise generate move value with up or down direction
               `;
             }
 
-            const nextMoveOptions: ExtractOptions<MoveAnalysis> = {
+            const nextMoveOptions: ExtractOptions<any> = {
               instruction: instructionStr,
-              schema: MoveAnalysis,
-              use_text_extract: false
+              schema: z.object({
+                move: z.number().optional(),
+                confidence: z.number().optional(),
+                reasoning: z.string().optional(),
+              }),
+              use_text_extract: false,
             };
 
-            const [moveSuccess, nextMove] = await session.browser.agent.extract(nextMoveOptions, page);
+            const [moveSuccess, nextMove] = await session.browser.agent.extract(
+              nextMoveOptions,
+              page
+            );
             let selectedMove = 4; // Default to no move
 
-            if (moveSuccess && nextMove.length > 0) {
-              selectedMove = nextMove[0].move ?? 4;
+            if (moveSuccess && nextMove && typeof nextMove === "object") {
+              selectedMove = (nextMove as any).move ?? 4;
             } else {
               console.log("❌ Failed to extract next move, retry observing");
               continue;

package/docs/examples/browser-use/browser/run-sudoku.ts

@@ -6,20 +6,28 @@
  * - Utilize PageUseAgent to run sudoku game
  */
 
-import { AgentBay, CreateSessionParams,BrowserOption, ExtractOptions, ActOptions } from 'wuying-agentbay-sdk';
-
-import { chromium } from 'playwright';
-
-class SudokuBoard {
-  board: number[][] = [];
-}
-
-class SudokuSolution {
-  solution: number[][] = [];
-}
-
+import {
+  AgentBay,
+  CreateSessionParams,
+  BrowserOption,
+  ExtractOptions,
+  ActOptions,
+} from "wuying-agentbay-sdk";
+
+import { chromium } from "playwright";
+import { z } from "zod";
+
+const SudokuBoardSchema = z.object({
+  board: z.array(z.array(z.number())),
+});
+type SudokuBoardType = z.infer<typeof SudokuBoardSchema>;
+
+const SudokuSolutionSchema = z.object({
+  solution: z.array(z.array(z.number())),
+});
+type SudokuSolutionType = z.infer<typeof SudokuSolutionSchema>;
 function formatBoardForLlm(board: number[][]): string {
-  return board.map(row => `  [${row.join(', ')}]`).join('\n');
+  return board.map((row) => `  [${row.join(", ")}]`).join("\n");
 }
 
 async function main() {
@@ -52,7 +60,7 @@ async function main() {
 
       const browser = await chromium.connectOverCDP(endpointUrl);
       try {
-        const context = browser.contexts()[0]
+        const context = browser.contexts()[0];
         const page = await context.newPage();
         console.log("🌐 Navigating to Sudoku site...");
         const url = "https://widget.websudoku.com/";
@@ -65,27 +73,31 @@ async function main() {
 
         while (!success) {
           console.log("📊 Extracting sudoku board...");
-          const options: ExtractOptions<SudokuBoard> = {
-            instruction: "Extract the current sudoku board as a 9x9 array. Each cell should be a number (1-9) if filled, or 0 if empty.",
-            schema: SudokuBoard,
-            use_text_extract: false
+          const options: ExtractOptions<typeof SudokuBoardSchema> = {
+            instruction:
+              "Extract the current sudoku board as a 9x9 array. Each cell should be a number (1-9) if filled, or 0 if empty.",
+            schema: SudokuBoardSchema,
+            use_text_extract: false,
           };
 
-          const [extractSuccess, boardObjs] = await session.browser.agent.extract(options, page);
-          if (extractSuccess && boardObjs.length > 0) {
+          const [extractSuccess, boardObj] =
+            await session.browser.agent.extract(options, page);
+          if (extractSuccess && boardObj) {
             success = true;
-            board = boardObjs[0].board;
+            board = boardObj.board;
           } else {
             console.log("❌ Failed to extract sudoku board, retry extracting");
-            await new Promise(resolve => setTimeout(resolve, 3000));
+            await new Promise((resolve) => setTimeout(resolve, 3000));
           }
         }
 
-        console.log("Current Board:\n" + board.map(row => row.join(' ')).join('\n'));
-        const originalBoard = board.map(row => [...row]);
+        console.log(
+          "Current Board:\n" + board.map((row) => row.join(" ")).join("\n")
+        );
+        const originalBoard = board.map((row) => [...row]);
 
         // 2. Solve the sudoku
-        const solutionOptions: ExtractOptions<SudokuSolution> = {
+        const solutionOptions: ExtractOptions<typeof SudokuSolutionSchema> = {
           instruction: `You are an expert sudoku solver. Given the following sudoku board as a 9x9 array (0 means empty), solve the sudoku and return the completed 9x9 array as 'solution'.
 
 Sudoku rules:
@@ -103,39 +115,46 @@ Return:
 {
   solution: number[][] // 9x9, all filled, valid sudoku
 }`,
-          schema: SudokuSolution,
-          use_text_extract: false
+          schema: SudokuSolutionSchema,
+          use_text_extract: false,
         };
 
-        const [solutionSuccess, solutionObjs] = await session.browser.agent.extract(solutionOptions, page);
-        if (!solutionSuccess || solutionObjs.length === 0) {
+        const [solutionSuccess, solutionObj] =
+          await session.browser.agent.extract(solutionOptions, page);
+        if (!solutionSuccess || !solutionObj) {
           console.log("❌ Failed to solve sudoku");
           return;
         }
 
-        const solution = solutionObjs[0].solution;
-        console.log("Solved Board:\n" + solution.map((row:any) => row.join(' ')).join('\n'));
+        const solution = solutionObj.solution;
+        console.log(
+          "Solved Board:\n" +
+            solution.map((row: any) => row.join(" ")).join("\n")
+        );
 
         // 3. Fill the solution
         for (let row = 0; row < 9; row++) {
           for (let col = 0; col < 9; col++) {
             if (originalBoard[row][col] === 0) {
               const inputId = `f${col}${row}`;
-              console.log(`Type '${solution[row][col]}' into the cell with id '${inputId}'`);
+              console.log(
+                `Type '${solution[row][col]}' into the cell with id '${inputId}'`
+              );
 
               // Use the act method for natural language action
               const actOptions: ActOptions = {
-                action: `Enter '${solution[row][col]}' into the input element where the attribute id is exactly '${inputId}' (for example, if id='f53', you must match the full string 'f53', not just the number 53; do not split or extract numbers from the id)`
+                action: `Enter '${solution[row][col]}' into the input element where the attribute id is exactly '${inputId}' (for example, if id='f53', you must match the full string 'f53', not just the number 53; do not split or extract numbers from the id)`,
               };
 
               await session.browser.agent.act(actOptions, page);
-              await new Promise(resolve => setTimeout(resolve, 500));
+              await new Promise((resolve) => setTimeout(resolve, 500));
             }
           }
         }
 
-        console.log("✅ Finished! The board has been solved and filled in the browser.");
-
+        console.log(
+          "✅ Finished! The board has been solved and filled in the browser."
+        );
       } catch (error) {
         console.log(`❌ Error in game loop: ${error}`);
       }

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "test-wuying-agentbay-sdk",
-  "version": "0.13.0-beta.20251212114304",
+  "version": "0.13.0-beta.20251212140302",
   "description": "TypeScript SDK for interacting with the Wuying AgentBay cloud runtime environment",
   "main": "./dist/index.cjs",
   "module": "./dist/index.mjs",
@@ -66,7 +66,9 @@
     "ts-jest": "^26.5.6",
     "tsup": "^8.5.0",
     "typedoc-plugin-markdown": "^3.16.0",
-    "typescript": "^4.9.5"
+    "typescript": "^4.9.5",
+    "zod": "^3.25.76",
+    "zod-to-json-schema": "^3.22.4"
   },
   "dependencies": {
     "@alicloud/openapi-core": "^1.0.4",