test-wuying-agentbay-sdk 0.13.1-beta.20251223203147 → 0.13.1-beta.20251224094729

package/dist/index.d.mts

@@ -3510,9 +3510,16 @@ interface ExecutionResult extends ApiResponse {
  * Result of query operations.
  */
 interface QueryResult extends ApiResponse {
+    taskId: string;
     taskStatus: string;
     taskAction: string;
     taskProduct: string;
+    stream?: Array<{
+        content?: string;
+        reasoning?: string;
+        timestamp_ms?: number;
+    }>;
+    error?: string;
 }
 /**
  * Result of agent initialization.
@@ -3546,94 +3553,38 @@ interface McpSession {
     callMcpTool(toolName: string, args: any): Promise<McpToolResult>;
 }
 /**
- * An Agent to perform tasks on the computer.
+ * Base class for task execution agents.
+ * Provides common functionality for ComputerUseAgent and BrowserUseAgent.
  */
-declare class ComputerUseAgent {
-    private session;
+declare abstract class BaseTaskAgent {
+    protected session: McpSession;
+    protected abstract toolPrefix: string;
+    constructor(session: McpSession);
     /**
-     * Initialize an Computer Agent object.
-     *
-     * @param session - The Session instance that this Agent belongs to.
+     * Get the full MCP tool name based on prefix and action.
      */
-    constructor(session: McpSession);
+    protected getToolName(action: string): string;
     /**
      * Execute a specific task described in human language.
-     *
-     * @param task - Task description in human language.
-     * @param maxTryTimes - Maximum number of retry attempts.
-     * @returns ExecutionResult containing success status, task output, and error
-     *     message if any.
-     *
-     * @example
-     * ```typescript
-     * const agentBay = new AgentBay({ apiKey: 'your_api_key' });
-     * const result = await agentBay.create({ imageId: 'windows_latest' });
-     * if (result.success) {
-     *   const taskResult = await result.session.agent.computer.executeTask(
-     *     'Open notepad',
-     *     10
-     *   );
-     *   console.log(`Task status: ${taskResult.taskStatus}`);
-     *   await result.session.delete();
-     * }
-     * ```
      */
     executeTask(task: string, maxTryTimes: number): Promise<ExecutionResult>;
     /**
      * Get the status of the task with the given task ID.
-     *
-     * @param taskId - Task ID
-     * @returns QueryResult containing the task status
-     *
-     * @example
-     * ```typescript
-     * const agentBay = new AgentBay({ apiKey: 'your_api_key' });
-     * const result = await agentBay.create({ imageId: 'windows_latest' });
-     * if (result.success) {
-     *   const taskResult = await result.session.agent.computer.executeTask('Open
-     * calculator', 10); const statusResult = await
-     * result.session.agent.computer.getTaskStatus(taskResult.taskId);
-     * console.log(`Status:
-     * ${JSON.parse(statusResult.output).status}`); await result.session.delete();
-     * }
-     * ```
      */
     getTaskStatus(taskId: string): Promise<QueryResult>;
     /**
      * Terminate a task with a specified task ID.
-     *
-     * @param taskId - The ID of the running task.
-     * @returns ExecutionResult containing success status, task output, and
-     *     error message if any.
-     *
-     * @example
-     * ```typescript
-     * const agentBay = new AgentBay({ apiKey: 'your_api_key' });
-     * const result = await agentBay.create({ imageId: 'windows_latest' });
-     * if (result.success) {
-     *   const taskResult = await result.session.agent.computer.executeTask(
-     *     'Open notepad',
-     *     5
-     *   );
-     *   const terminateResult = await result.session.agent.computer.terminateTask(
-     *     taskResult.taskId
-     *   );
-     *   console.log(`Terminated: ${terminateResult.taskStatus}`);
-     *   await result.session.delete();
-     * }
-     * ```
      */
     terminateTask(taskId: string): Promise<ExecutionResult>;
 }
-declare class BrowserUseAgent {
-    private session;
-    /**
-     * Initialize an Browser Agent object.
-     * @description Browser Use Agent is in BETA ⚠️ .
-     *
-     * @param session - The Session instance that this Agent belongs to.
-     */
-    constructor(session: McpSession);
+/**
+ * An Agent to perform tasks on the computer.
+ */
+declare class ComputerUseAgent extends BaseTaskAgent {
+    protected toolPrefix: string;
+}
+declare class BrowserUseAgent extends BaseTaskAgent {
+    protected toolPrefix: string;
     /**
      * Initialize the browser agent with specific options.
      * @param options - agent initialization options
@@ -3653,78 +3604,74 @@ declare class BrowserUseAgent {
      * ```
      */
     initialize(options: AgentOptions): Promise<InitializationResult>;
+}
+/**
+ * An Agent to perform tasks on mobile devices.
+ */
+declare class MobileUseAgent extends BaseTaskAgent {
+    protected toolPrefix: string;
     /**
-     * Execute a specific task described in human language.
+     * Execute a task in human language without waiting for completion
+     * (non-blocking). This is a fire-and-return interface that immediately
+     * provides a task ID. Call getTaskStatus to check the task status.
      *
      * @param task - Task description in human language.
-     * @param maxTryTimes - Maximum number of retry attempts.
-     * @returns ExecutionResult containing success status, task output, and
-     *     error message if any.
-     *
-     * @example
-     * ```typescript
-     * const agentBay = new AgentBay({ apiKey: 'your_api_key' });
-     * const result = await agentBay.create({ imageId: 'linux_latest' });
-     * if (result.success) {
-     *   const taskResult = await result.session.agent.browser.executeTask(
-     *     'Navigate to baidu and query the weather of Shanghai',
-     *     10
-     *   );
-     *   console.log(`Task status: ${taskResult.taskStatus}`);
-     *   await result.session.delete();
-     * }
-     * ```
-     */
-    executeTask(task: string, maxTryTimes: number): Promise<ExecutionResult>;
-    /**
-     * Get the status of the task with the given task ID.
-     *
-     * @param taskId - Task ID
-     * @returns QueryResult containing the task status
+     * @param maxSteps - Maximum number of steps (clicks/swipes/etc.) allowed.
+     *                   Used to prevent infinite loops or excessive resource
+     *                   consumption. Default is 50.
+     * @param maxStepRetries - Maximum retry times for MCP tool call failures
+     *                         at SDK level. Used to retry when callMcpTool fails
+     *                         (e.g., network errors, timeouts). Default is 3.
+     * @returns ExecutionResult containing success status, task ID, task status,
+     *     and error message if any.
      *
      * @example
      * ```typescript
      * const agentBay = new AgentBay({ apiKey: 'your_api_key' });
-     * const result = await agentBay.create({ imageId: 'linux_latest' });
+     * const result = await agentBay.create({ imageId: 'mobile_latest' });
      * if (result.success) {
-     *   const taskResult = await result.session.agent.browser.executeTask(
-     *     'Navigate to baidu and query the weather of Shanghai',
-     *     10
-     *   );
-     *   const statusResult = await result.session.agent.browser.getTaskStatus(
-     *     taskResult.taskId
+     *   const execResult = await result.session.agent.mobile.executeTask(
+     *     'Open WeChat app', 100, 5
      *   );
-     *   console.log(`Status: ${statusResult.taskStatus}`);
+     *   console.log(`Task ID: ${execResult.taskId}`);
      *   await result.session.delete();
      * }
      * ```
      */
-    getTaskStatus(taskId: string): Promise<QueryResult>;
+    executeTask(task: string, maxSteps?: number, maxStepRetries?: number): Promise<ExecutionResult>;
     /**
-     * Terminate a task with a specified task ID.
+     * Execute a specific task described in human language synchronously.
+     * This is a synchronous interface that blocks until the task is completed or
+     * an error occurs, or timeout happens. The default polling interval is
+     * 3 seconds.
      *
-     * @param taskId - The ID of the running task.
-     * @returns ExecutionResult containing success status, task output, and
-     *     error message if any.
+     * @param task - Task description in human language.
+     * @param maxSteps - Maximum number of steps (clicks/swipes/etc.) allowed.
+     *                   Used to prevent infinite loops or excessive resource
+     *                   consumption. Default is 50.
+     * @param maxStepRetries - Maximum retry times for MCP tool call failures
+     *                         at SDK level. Used to retry when callMcpTool fails
+     *                         (e.g., network errors, timeouts). Default is 3.
+     * @param maxTryTimes - Maximum number of polling attempts (each 3 seconds).
+     *                      Used to control how long to wait for task completion.
+     *                      Default is 300 (about 15 minutes).
+     * @returns ExecutionResult containing success status, task ID, task status,
+     *     and error message if any.
      *
      * @example
      * ```typescript
      * const agentBay = new AgentBay({ apiKey: 'your_api_key' });
-     * const result = await agentBay.create({ imageId: 'linux_latest' });
+     * const result = await agentBay.create({ imageId: 'mobile_latest' });
      * if (result.success) {
-     *   const taskResult = await result.session.agent.browser.executeTask(
-     *     'Navigate to baidu and query the weather of Shanghai',
-     *     10
-     *   );
-     *   const terminateResult = await result.session.agent.browser.terminateTask(
-     *     taskResult.taskId
+     *   const execResult = await result.session.agent.mobile.executeTaskAndWait(
+     *     'Open WeChat app', 100, 3, 200
      *   );
-     *   console.log(`Terminated: ${terminateResult.taskStatus}`);
+     *   console.log(`Task result: ${execResult.taskResult}`);
      *   await result.session.delete();
      * }
      * ```
      */
-    terminateTask(taskId: string): Promise<ExecutionResult>;
+    executeTaskAndWait(task: string, maxSteps?: number, maxStepRetries?: number, maxTryTimes?: number): Promise<ExecutionResult>;
 }
 /**
  * An Agent to manipulate applications to complete specific tasks.
@@ -3741,6 +3688,10 @@ declare class Agent {
      * An instance of Browser Use Agent.
      */
     browser: BrowserUseAgent;
+    /**
+     * An instance of Mobile Use Agent.
+     */
+    mobile: MobileUseAgent;
     /**
      * Initialize an Agent object.
      *
@@ -7610,6 +7561,10 @@ declare class MobileSimulateService {
  * Log level type
  */
 type LogLevel = 'DEBUG' | 'INFO' | 'WARN' | 'ERROR';
+/**
+ * Log format type
+ */
+type LogFormat = 'pretty' | 'sls';
 /**
  * Logger configuration options
  *
@@ -7649,6 +7604,7 @@ interface LoggerConfig {
     logFile?: string;
     maxFileSize?: string;
     enableConsole?: boolean;
+    format?: LogFormat | string;
 }
 /**
  * Set the log level

package/dist/index.d.ts

@@ -3510,9 +3510,16 @@ interface ExecutionResult extends ApiResponse {
  * Result of query operations.
  */
 interface QueryResult extends ApiResponse {
+    taskId: string;
     taskStatus: string;
     taskAction: string;
     taskProduct: string;
+    stream?: Array<{
+        content?: string;
+        reasoning?: string;
+        timestamp_ms?: number;
+    }>;
+    error?: string;
 }
 /**
  * Result of agent initialization.
@@ -3546,94 +3553,38 @@ interface McpSession {
     callMcpTool(toolName: string, args: any): Promise<McpToolResult>;
 }
 /**
- * An Agent to perform tasks on the computer.
+ * Base class for task execution agents.
+ * Provides common functionality for ComputerUseAgent and BrowserUseAgent.
  */
-declare class ComputerUseAgent {
-    private session;
+declare abstract class BaseTaskAgent {
+    protected session: McpSession;
+    protected abstract toolPrefix: string;
+    constructor(session: McpSession);
     /**
-     * Initialize an Computer Agent object.
-     *
-     * @param session - The Session instance that this Agent belongs to.
+     * Get the full MCP tool name based on prefix and action.
      */
-    constructor(session: McpSession);
+    protected getToolName(action: string): string;
     /**
      * Execute a specific task described in human language.
-     *
-     * @param task - Task description in human language.
-     * @param maxTryTimes - Maximum number of retry attempts.
-     * @returns ExecutionResult containing success status, task output, and error
-     *     message if any.
-     *
-     * @example
-     * ```typescript
-     * const agentBay = new AgentBay({ apiKey: 'your_api_key' });
-     * const result = await agentBay.create({ imageId: 'windows_latest' });
-     * if (result.success) {
-     *   const taskResult = await result.session.agent.computer.executeTask(
-     *     'Open notepad',
-     *     10
-     *   );
-     *   console.log(`Task status: ${taskResult.taskStatus}`);
-     *   await result.session.delete();
-     * }
-     * ```
      */
     executeTask(task: string, maxTryTimes: number): Promise<ExecutionResult>;
     /**
      * Get the status of the task with the given task ID.
-     *
-     * @param taskId - Task ID
-     * @returns QueryResult containing the task status
-     *
-     * @example
-     * ```typescript
-     * const agentBay = new AgentBay({ apiKey: 'your_api_key' });
-     * const result = await agentBay.create({ imageId: 'windows_latest' });
-     * if (result.success) {
-     *   const taskResult = await result.session.agent.computer.executeTask('Open
-     * calculator', 10); const statusResult = await
-     * result.session.agent.computer.getTaskStatus(taskResult.taskId);
-     * console.log(`Status:
-     * ${JSON.parse(statusResult.output).status}`); await result.session.delete();
-     * }
-     * ```
      */
     getTaskStatus(taskId: string): Promise<QueryResult>;
     /**
      * Terminate a task with a specified task ID.
-     *
-     * @param taskId - The ID of the running task.
-     * @returns ExecutionResult containing success status, task output, and
-     *     error message if any.
-     *
-     * @example
-     * ```typescript
-     * const agentBay = new AgentBay({ apiKey: 'your_api_key' });
-     * const result = await agentBay.create({ imageId: 'windows_latest' });
-     * if (result.success) {
-     *   const taskResult = await result.session.agent.computer.executeTask(
-     *     'Open notepad',
-     *     5
-     *   );
-     *   const terminateResult = await result.session.agent.computer.terminateTask(
-     *     taskResult.taskId
-     *   );
-     *   console.log(`Terminated: ${terminateResult.taskStatus}`);
-     *   await result.session.delete();
-     * }
-     * ```
      */
     terminateTask(taskId: string): Promise<ExecutionResult>;
 }
-declare class BrowserUseAgent {
-    private session;
-    /**
-     * Initialize an Browser Agent object.
-     * @description Browser Use Agent is in BETA ⚠️ .
-     *
-     * @param session - The Session instance that this Agent belongs to.
-     */
-    constructor(session: McpSession);
+/**
+ * An Agent to perform tasks on the computer.
+ */
+declare class ComputerUseAgent extends BaseTaskAgent {
+    protected toolPrefix: string;
+}
+declare class BrowserUseAgent extends BaseTaskAgent {
+    protected toolPrefix: string;
     /**
      * Initialize the browser agent with specific options.
      * @param options - agent initialization options
@@ -3653,78 +3604,74 @@ declare class BrowserUseAgent {
      * ```
      */
     initialize(options: AgentOptions): Promise<InitializationResult>;
+}
+/**
+ * An Agent to perform tasks on mobile devices.
+ */
+declare class MobileUseAgent extends BaseTaskAgent {
+    protected toolPrefix: string;
     /**
-     * Execute a specific task described in human language.
+     * Execute a task in human language without waiting for completion
+     * (non-blocking). This is a fire-and-return interface that immediately
+     * provides a task ID. Call getTaskStatus to check the task status.
      *
      * @param task - Task description in human language.
-     * @param maxTryTimes - Maximum number of retry attempts.
-     * @returns ExecutionResult containing success status, task output, and
-     *     error message if any.
-     *
-     * @example
-     * ```typescript
-     * const agentBay = new AgentBay({ apiKey: 'your_api_key' });
-     * const result = await agentBay.create({ imageId: 'linux_latest' });
-     * if (result.success) {
-     *   const taskResult = await result.session.agent.browser.executeTask(
-     *     'Navigate to baidu and query the weather of Shanghai',
-     *     10
-     *   );
-     *   console.log(`Task status: ${taskResult.taskStatus}`);
-     *   await result.session.delete();
-     * }
-     * ```
-     */
-    executeTask(task: string, maxTryTimes: number): Promise<ExecutionResult>;
-    /**
-     * Get the status of the task with the given task ID.
-     *
-     * @param taskId - Task ID
-     * @returns QueryResult containing the task status
+     * @param maxSteps - Maximum number of steps (clicks/swipes/etc.) allowed.
+     *                   Used to prevent infinite loops or excessive resource
+     *                   consumption. Default is 50.
+     * @param maxStepRetries - Maximum retry times for MCP tool call failures
+     *                         at SDK level. Used to retry when callMcpTool fails
+     *                         (e.g., network errors, timeouts). Default is 3.
+     * @returns ExecutionResult containing success status, task ID, task status,
+     *     and error message if any.
      *
      * @example
      * ```typescript
      * const agentBay = new AgentBay({ apiKey: 'your_api_key' });
-     * const result = await agentBay.create({ imageId: 'linux_latest' });
+     * const result = await agentBay.create({ imageId: 'mobile_latest' });
      * if (result.success) {
-     *   const taskResult = await result.session.agent.browser.executeTask(
-     *     'Navigate to baidu and query the weather of Shanghai',
-     *     10
-     *   );
-     *   const statusResult = await result.session.agent.browser.getTaskStatus(
-     *     taskResult.taskId
+     *   const execResult = await result.session.agent.mobile.executeTask(
+     *     'Open WeChat app', 100, 5
      *   );
-     *   console.log(`Status: ${statusResult.taskStatus}`);
+     *   console.log(`Task ID: ${execResult.taskId}`);
      *   await result.session.delete();
      * }
      * ```
      */
-    getTaskStatus(taskId: string): Promise<QueryResult>;
+    executeTask(task: string, maxSteps?: number, maxStepRetries?: number): Promise<ExecutionResult>;
     /**
-     * Terminate a task with a specified task ID.
+     * Execute a specific task described in human language synchronously.
+     * This is a synchronous interface that blocks until the task is completed or
+     * an error occurs, or timeout happens. The default polling interval is
+     * 3 seconds.
      *
-     * @param taskId - The ID of the running task.
-     * @returns ExecutionResult containing success status, task output, and
-     *     error message if any.
+     * @param task - Task description in human language.
+     * @param maxSteps - Maximum number of steps (clicks/swipes/etc.) allowed.
+     *                   Used to prevent infinite loops or excessive resource
+     *                   consumption. Default is 50.
+     * @param maxStepRetries - Maximum retry times for MCP tool call failures
+     *                         at SDK level. Used to retry when callMcpTool fails
+     *                         (e.g., network errors, timeouts). Default is 3.
+     * @param maxTryTimes - Maximum number of polling attempts (each 3 seconds).
+     *                      Used to control how long to wait for task completion.
+     *                      Default is 300 (about 15 minutes).
+     * @returns ExecutionResult containing success status, task ID, task status,
+     *     and error message if any.
      *
      * @example
      * ```typescript
      * const agentBay = new AgentBay({ apiKey: 'your_api_key' });
-     * const result = await agentBay.create({ imageId: 'linux_latest' });
+     * const result = await agentBay.create({ imageId: 'mobile_latest' });
      * if (result.success) {
-     *   const taskResult = await result.session.agent.browser.executeTask(
-     *     'Navigate to baidu and query the weather of Shanghai',
-     *     10
-     *   );
-     *   const terminateResult = await result.session.agent.browser.terminateTask(
-     *     taskResult.taskId
+     *   const execResult = await result.session.agent.mobile.executeTaskAndWait(
+     *     'Open WeChat app', 100, 3, 200
      *   );
-     *   console.log(`Terminated: ${terminateResult.taskStatus}`);
+     *   console.log(`Task result: ${execResult.taskResult}`);
      *   await result.session.delete();
      * }
      * ```
      */
-    terminateTask(taskId: string): Promise<ExecutionResult>;
+    executeTaskAndWait(task: string, maxSteps?: number, maxStepRetries?: number, maxTryTimes?: number): Promise<ExecutionResult>;
 }
 /**
  * An Agent to manipulate applications to complete specific tasks.
@@ -3741,6 +3688,10 @@ declare class Agent {
      * An instance of Browser Use Agent.
      */
     browser: BrowserUseAgent;
+    /**
+     * An instance of Mobile Use Agent.
+     */
+    mobile: MobileUseAgent;
     /**
      * Initialize an Agent object.
      *
@@ -7610,6 +7561,10 @@ declare class MobileSimulateService {
  * Log level type
  */
 type LogLevel = 'DEBUG' | 'INFO' | 'WARN' | 'ERROR';
+/**
+ * Log format type
+ */
+type LogFormat = 'pretty' | 'sls';
 /**
  * Logger configuration options
  *
@@ -7649,6 +7604,7 @@ interface LoggerConfig {
     logFile?: string;
     maxFileSize?: string;
     enableConsole?: boolean;
+    format?: LogFormat | string;
 }
 /**
  * Set the log level