page-agent 0.0.0 → 0.0.1

package/dist/lib/page-agent.js

@@ -22,19 +22,18 @@ var __privateGet = (obj, member, getter) => (__accessCheck(obj, member, "read fr
 var __privateAdd = (obj, member, value) => member.has(obj) ? __typeError("Cannot add the same private member more than once") : member instanceof WeakSet ? member.add(obj) : member.set(obj, value);
 var __privateSet = (obj, member, value, setter) => (__accessCheck(obj, member, "write to private field"), setter ? setter.call(obj, value) : member.set(obj, value), value);
 var __privateMethod = (obj, member, method) => (__accessCheck(obj, member, "access private method"), method);
-var _openai, _model, _bus, _wrapper, _indicator, _statusText, _historySection, _expandButton, _pauseButton, _stopButton, _inputSection, _taskInput, _bus2, _state, _isExpanded, _pageAgent, _userAnswerResolver, _isWaitingForUserAnswer, _Panel_instances, update_fn, show_fn, hide_fn, reset_fn, togglePause_fn, updatePauseButton_fn, stopAgent_fn, submitTask_fn, handleUserAnswer_fn, showInputArea_fn, hideInputArea_fn, shouldShowInputArea_fn, createWrapper_fn, setupEventListeners_fn, toggle_fn, expand_fn, collapse_fn, animateTextChange_fn, updateStatusIndicator_fn, updateHistory_fn, scrollToBottom_fn, createHistoryItem_fn, _cursor, _currentCursorX, _currentCursorY, _targetCursorX, _targetCursorY, _SimulatorMask_instances, createCursor_fn, moveCursorToTarget_fn, _llm, _totalWaitTime, _abortController, _PageAgent_instances, packMacroTool_fn, getSystemPrompt_fn, assembleUserPrompt_fn, onDone_fn, getBrowserState_fn, updateTree_fn;
-import { generateText, stepCountIs, tool } from "ai";
+var _bus, _wrapper, _indicator, _statusText, _historySection, _expandButton, _pauseButton, _stopButton, _inputSection, _taskInput, _bus2, _state, _isExpanded, _pageAgent, _userAnswerResolver, _isWaitingForUserAnswer, _Panel_instances, update_fn, show_fn, hide_fn, reset_fn, togglePause_fn, updatePauseButton_fn, stopAgent_fn, submitTask_fn, handleUserAnswer_fn, showInputArea_fn, hideInputArea_fn, shouldShowInputArea_fn, createWrapper_fn, setupEventListeners_fn, toggle_fn, expand_fn, collapse_fn, animateTextChange_fn, updateStatusIndicator_fn, updateHistory_fn, scrollToBottom_fn, createHistoryItem_fn, _cursor, _currentCursorX, _currentCursorY, _targetCursorX, _targetCursorY, _SimulatorMask_instances, createCursor_fn, moveCursorToTarget_fn, _llm, _totalWaitTime, _abortController, _PageAgent_instances, packMacroTool_fn, getSystemPrompt_fn, assembleUserPrompt_fn, onDone_fn, getBrowserState_fn, updateTree_fn;
 import chalk from "chalk";
-import zod from "zod";
-import { createOpenAI } from "@ai-sdk/openai";
+import zod, { z } from "zod";
 import { Motion } from "ai-motion";
 const VIEWPORT_EXPANSION = -1;
 const DEFAULT_MODEL_NAME = "gpt-41-mini-0414-global";
 const DEFAULT_API_KEY = "not-needed";
 const DEFAULT_BASE_URL = "http://localhost:3000/api/agent";
-const MACRO_TOOL_NAME = "AgentOutput";
 const LLM_MAX_RETRIES = 2;
 const MAX_STEPS = 20;
+const DEFAULT_TEMPERATURE = 0.7;
+const DEFAULT_MAX_TOKENS = 4096;
 const domTree = /* @__PURE__ */ __name((args = {
   doHighlightElements: true,
   focusHighlightIndex: -1,
@@ -1396,49 +1395,6 @@ function getPageInfo() {
   };
 }
 __name(getPageInfo, "getPageInfo");
-const zhCN = {
-  ui: {
-    panel: {
-      ready: "准备就绪",
-      thinking: "正在思考...",
-      paused: "暂停中，稍后",
-      taskInput: "输入新任务，详细描述步骤，回车提交",
-      userAnswerPrompt: "请回答上面问题，回车提交",
-      taskTerminated: "任务已终止",
-      taskCompleted: "任务结束",
-      continueExecution: "继续执行",
-      userAnswer: "用户回答: {{input}}",
-      pause: "暂停",
-      continue: "继续",
-      stop: "终止",
-      expand: "展开历史",
-      collapse: "收起历史",
-      step: "步骤 {{number}} · {{time}}{{duration}}"
-    },
-    tools: {
-      clicking: "正在点击元素 [{{index}}]...",
-      inputting: "正在输入文本到元素 [{{index}}]...",
-      selecting: '正在选择选项 "{{text}}"...',
-      scrolling: "正在滚动页面...",
-      waiting: "等待 {{seconds}} 秒...",
-      done: "结束任务",
-      clicked: "🖱️ 已点击元素 [{{index}}]",
-      inputted: '⌨️ 已输入文本 "{{text}}"',
-      selected: '☑️ 已选择选项 "{{text}}"',
-      scrolled: "🛞 页面滚动完成",
-      waited: "⌛️ 等待完成",
-      executing: "正在执行 {{toolName}}..."
-    },
-    errors: {
-      elementNotFound: "未找到索引为 {{index}} 的交互元素",
-      taskRequired: "任务描述不能为空",
-      executionFailed: "任务执行失败",
-      notInputElement: "元素不是输入框或文本域",
-      notSelectElement: "元素不是选择框",
-      optionNotFound: '未找到选项 "{{text}}"'
-    }
-  }
-};
 const enUS = {
   ui: {
     panel: {
@@ -1451,6 +1407,8 @@ const enUS = {
       taskCompleted: "Task completed",
       continueExecution: "Continue execution",
       userAnswer: "User answer: {{input}}",
+      question: "Question: {{question}}",
+      waitingPlaceholder: "Waiting for task to start...",
       pause: "Pause",
       continue: "Continue",
       stop: "Stop",
@@ -1470,7 +1428,10 @@ const enUS = {
       selected: '☑️ Selected option "{{text}}"',
       scrolled: "🛞 Page scrolled",
       waited: "⌛️ Wait completed",
-      executing: "正在执行 {{toolName}}..."
+      executing: "Executing {{toolName}}...",
+      resultSuccess: "success",
+      resultFailure: "failed",
+      resultError: "error"
     },
     errors: {
       elementNotFound: "No interactive element found at index {{index}}",
@@ -1482,9 +1443,57 @@ const enUS = {
     }
   }
 };
+const zhCN = {
+  ui: {
+    panel: {
+      ready: "准备就绪",
+      thinking: "正在思考...",
+      paused: "暂停中，稍后",
+      taskInput: "输入新任务，详细描述步骤，回车提交",
+      userAnswerPrompt: "请回答上面问题，回车提交",
+      taskTerminated: "任务已终止",
+      taskCompleted: "任务结束",
+      continueExecution: "继续执行",
+      userAnswer: "用户回答: {{input}}",
+      question: "询问: {{question}}",
+      waitingPlaceholder: "等待任务开始...",
+      pause: "暂停",
+      continue: "继续",
+      stop: "终止",
+      expand: "展开历史",
+      collapse: "收起历史",
+      step: "步骤 {{number}} · {{time}}{{duration}}"
+    },
+    tools: {
+      clicking: "正在点击元素 [{{index}}]...",
+      inputting: "正在输入文本到元素 [{{index}}]...",
+      selecting: '正在选择选项 "{{text}}"...',
+      scrolling: "正在滚动页面...",
+      waiting: "等待 {{seconds}} 秒...",
+      done: "结束任务",
+      clicked: "🖱️ 已点击元素 [{{index}}]",
+      inputted: '⌨️ 已输入文本 "{{text}}"',
+      selected: '☑️ 已选择选项 "{{text}}"',
+      scrolled: "🛞 页面滚动完成",
+      waited: "⌛️ 等待完成",
+      executing: "正在执行 {{toolName}}...",
+      resultSuccess: "成功",
+      resultFailure: "失败",
+      resultError: "错误"
+    },
+    errors: {
+      elementNotFound: "未找到索引为 {{index}} 的交互元素",
+      taskRequired: "任务描述不能为空",
+      executionFailed: "任务执行失败",
+      notInputElement: "元素不是输入框或文本域",
+      notSelectElement: "元素不是选择框",
+      optionNotFound: '未找到选项 "{{text}}"'
+    }
+  }
+};
 const locales = {
-  "zh-CN": zhCN,
-  "en-US": enUS
+  "en-US": enUS,
+  "zh-CN": zhCN
 };
 const _I18n = class _I18n {
   language;
@@ -1510,7 +1519,7 @@ const _I18n = class _I18n {
   }
   interpolate(template, params) {
     return template.replace(/\{\{(\w+)\}\}/g, (match, key) => {
-      return params[key]?.toString() || match;
+      return params[key] != null ? params[key].toString() : match;
     });
   }
   getLanguage() {
@@ -1519,14 +1528,17 @@ const _I18n = class _I18n {
 };
 __name(_I18n, "I18n");
 let I18n = _I18n;
-function assert(condition, message, silent) {
-  if (!condition) {
-    const errorMessage = message ?? "Assertion failed";
-    if (!silent) console.error(chalk.red(`❌ assert: ${errorMessage}`));
-    throw new Error(errorMessage);
-  }
+function parseLLMConfig(config) {
+  return {
+    baseURL: config.baseURL ?? DEFAULT_BASE_URL,
+    apiKey: config.apiKey ?? DEFAULT_API_KEY,
+    model: config.model ?? DEFAULT_MODEL_NAME,
+    temperature: config.temperature ?? DEFAULT_TEMPERATURE,
+    maxTokens: config.maxTokens ?? DEFAULT_MAX_TOKENS,
+    maxRetries: config.maxRetries ?? LLM_MAX_RETRIES
+  };
 }
-__name(assert, "assert");
+__name(parseLLMConfig, "parseLLMConfig");
 const _EventBus = class _EventBus extends EventTarget {
   /**
    * Listen to built-in events
@@ -1571,24 +1583,316 @@ function getEventBus(channel) {
   return bus;
 }
 __name(getEventBus, "getEventBus");
+const InvokeErrorType = {
+  // Retryable
+  NETWORK_ERROR: "network_error",
+  // Network error, retry
+  RATE_LIMIT: "rate_limit",
+  // Rate limit, retry
+  SERVER_ERROR: "server_error",
+  // 5xx, retry
+  NO_TOOL_CALL: "no_tool_call",
+  // Model did not call tool
+  INVALID_TOOL_ARGS: "invalid_tool_args",
+  // Tool args don't match schema
+  TOOL_EXECUTION_ERROR: "tool_execution_error",
+  // Tool execution error
+  UNKNOWN: "unknown",
+  // Non-retryable
+  AUTH_ERROR: "auth_error",
+  // Authentication failed
+  CONTEXT_LENGTH: "context_length",
+  // Prompt too long
+  CONTENT_FILTER: "content_filter"
+  // Content filtered
+};
+const _InvokeError = class _InvokeError extends Error {
+  type;
+  retryable;
+  statusCode;
+  rawError;
+  constructor(type, message, rawError) {
+    super(message);
+    this.name = "InvokeError";
+    this.type = type;
+    this.retryable = this.isRetryable(type);
+    this.rawError = rawError;
+  }
+  isRetryable(type) {
+    const retryableTypes = [
+      InvokeErrorType.NETWORK_ERROR,
+      InvokeErrorType.RATE_LIMIT,
+      InvokeErrorType.SERVER_ERROR,
+      InvokeErrorType.NO_TOOL_CALL,
+      InvokeErrorType.INVALID_TOOL_ARGS,
+      InvokeErrorType.TOOL_EXECUTION_ERROR,
+      InvokeErrorType.UNKNOWN
+    ];
+    return retryableTypes.includes(type);
+  }
+};
+__name(_InvokeError, "InvokeError");
+let InvokeError = _InvokeError;
+function zodToOpenAITool(name, tool2) {
+  return {
+    type: "function",
+    function: {
+      name,
+      description: tool2.description,
+      parameters: z.toJSONSchema(tool2.inputSchema, { target: "openapi-3.0" })
+    }
+  };
+}
+__name(zodToOpenAITool, "zodToOpenAITool");
+function lenientParseMacroToolCall(responseData, inputSchema) {
+  const choice = responseData.choices?.[0];
+  if (!choice) {
+    throw new InvokeError(InvokeErrorType.UNKNOWN, "No choices in response", responseData);
+  }
+  switch (choice.finish_reason) {
+    case "tool_calls":
+    case "stop":
+      break;
+    case "length":
+      throw new InvokeError(
+        InvokeErrorType.CONTEXT_LENGTH,
+        "Response truncated: max tokens reached"
+      );
+    case "content_filter":
+      throw new InvokeError(InvokeErrorType.CONTENT_FILTER, "Content filtered by safety system");
+    default:
+      throw new InvokeError(
+        InvokeErrorType.UNKNOWN,
+        `Unexpected finish_reason: ${choice.finish_reason}`
+      );
+  }
+  const actionSchema = inputSchema.shape.action;
+  if (!actionSchema) {
+    throw new Error('inputSchema must have an "action" field');
+  }
+  let arg = null;
+  const toolCall = choice.message?.tool_calls?.[0]?.function;
+  arg = toolCall?.arguments ?? null;
+  if (arg && toolCall.name !== "AgentOutput") {
+    console.log(chalk.yellow("lenientParseMacroToolCall: #1 fixing incorrect tool call"));
+    let tmpArg;
+    try {
+      tmpArg = JSON.parse(arg);
+    } catch (error2) {
+      throw new InvokeError(
+        InvokeErrorType.INVALID_TOOL_ARGS,
+        "Failed to parse tool arguments as JSON",
+        error2
+      );
+    }
+    arg = JSON.stringify({ action: { [toolCall.name]: tmpArg } });
+  }
+  if (!arg) {
+    arg = choice.message?.content.trim() || null;
+  }
+  if (!arg) {
+    throw new InvokeError(
+      InvokeErrorType.NO_TOOL_CALL,
+      "No tool call or content found in response",
+      responseData
+    );
+  }
+  let parsedArgs;
+  try {
+    parsedArgs = JSON.parse(arg);
+  } catch (error2) {
+    throw new InvokeError(
+      InvokeErrorType.INVALID_TOOL_ARGS,
+      "Failed to parse tool arguments as JSON",
+      error2
+    );
+  }
+  if (parsedArgs.action || parsedArgs.evaluation_previous_goal || parsedArgs.next_goal) {
+    if (!parsedArgs.action) {
+      console.log(chalk.yellow("lenientParseMacroToolCall: #2 fixing incorrect tool call"));
+      parsedArgs.action = {
+        wait: { seconds: 1 }
+      };
+    }
+  } else if (parsedArgs.type && parsedArgs.function) {
+    if (parsedArgs.function.name !== "AgentOutput")
+      throw new InvokeError(
+        InvokeErrorType.INVALID_TOOL_ARGS,
+        `Expected function name "AgentOutput", got "${parsedArgs.function.name}"`,
+        null
+      );
+    console.log(chalk.yellow("lenientParseMacroToolCall: #3 fixing incorrect tool call"));
+    parsedArgs = parsedArgs.function.arguments;
+  } else if (parsedArgs.name && parsedArgs.arguments) {
+    if (parsedArgs.name !== "AgentOutput")
+      throw new InvokeError(
+        InvokeErrorType.INVALID_TOOL_ARGS,
+        `Expected function name "AgentOutput", got "${parsedArgs.name}"`,
+        null
+      );
+    console.log(chalk.yellow("lenientParseMacroToolCall: #4 fixing incorrect tool call"));
+    parsedArgs = parsedArgs.arguments;
+  } else {
+    console.log(chalk.yellow("lenientParseMacroToolCall: #5 fixing incorrect tool call"));
+    parsedArgs = { action: parsedArgs };
+  }
+  if (typeof parsedArgs === "string") {
+    console.log(chalk.yellow("lenientParseMacroToolCall: #6 fixing incorrect tool call"));
+    try {
+      parsedArgs = JSON.parse(parsedArgs);
+    } catch (error2) {
+      throw new InvokeError(
+        InvokeErrorType.INVALID_TOOL_ARGS,
+        "Failed to parse nested tool arguments as JSON",
+        error2
+      );
+    }
+  }
+  const validation = inputSchema.safeParse(parsedArgs);
+  if (validation.success) {
+    return validation.data;
+  } else {
+    const action = parsedArgs.action ?? {};
+    const actionName = Object.keys(action)[0] || "unknown";
+    const actionArgs = JSON.stringify(action[actionName] || "unknown");
+    throw new InvokeError(
+      InvokeErrorType.INVALID_TOOL_ARGS,
+      `Tool arguments validation failed: action "${actionName}" with args ${actionArgs}`,
+      validation.error
+    );
+  }
+}
+__name(lenientParseMacroToolCall, "lenientParseMacroToolCall");
+function modelPatch(body) {
+  const model = body.model || "";
+  if (model.toLowerCase().startsWith("claude")) {
+    body.tool_choice = { type: "tool", name: "AgentOutput" };
+    body.thinking = { type: "disabled" };
+  }
+  if (model.toLowerCase().includes("grok")) {
+    console.log("Applying Grok patch: removing tool_choice");
+    delete body.tool_choice;
+    console.log("Applying Grok patch: disable reasoning and thinking");
+    body.thinking = { type: "disabled", effort: "minimal" };
+    body.reasoning = { enabled: false, effort: "low" };
+  }
+  return body;
+}
+__name(modelPatch, "modelPatch");
+const _OpenAIClient = class _OpenAIClient {
+  config;
+  constructor(config) {
+    this.config = config;
+  }
+  async invoke(messages, tools2, abortSignal) {
+    const openaiTools = Object.entries(tools2).map(([name, tool22]) => zodToOpenAITool(name, tool22));
+    let response;
+    try {
+      response = await fetch(`${this.config.baseURL}/chat/completions`, {
+        method: "POST",
+        headers: {
+          "Content-Type": "application/json",
+          Authorization: `Bearer ${this.config.apiKey}`
+        },
+        body: JSON.stringify(
+          modelPatch({
+            model: this.config.model,
+            temperature: this.config.temperature,
+            max_tokens: this.config.maxTokens,
+            messages,
+            tools: openaiTools,
+            // tool_choice: 'required',
+            tool_choice: { type: "function", function: { name: "AgentOutput" } },
+            // model specific params
+            // reasoning_effort: 'minimal',
+            // verbosity: 'low',
+            parallel_tool_calls: false
+          })
+        ),
+        signal: abortSignal
+      });
+    } catch (error2) {
+      throw new InvokeError(InvokeErrorType.NETWORK_ERROR, "Network request failed", error2);
+    }
+    if (!response.ok) {
+      const errorData = await response.json().catch();
+      const errorMessage = errorData.error?.message || response.statusText;
+      if (response.status === 401 || response.status === 403) {
+        throw new InvokeError(
+          InvokeErrorType.AUTH_ERROR,
+          `Authentication failed: ${errorMessage}`,
+          errorData
+        );
+      }
+      if (response.status === 429) {
+        throw new InvokeError(
+          InvokeErrorType.RATE_LIMIT,
+          `Rate limit exceeded: ${errorMessage}`,
+          errorData
+        );
+      }
+      if (response.status >= 500) {
+        throw new InvokeError(
+          InvokeErrorType.SERVER_ERROR,
+          `Server error: ${errorMessage}`,
+          errorData
+        );
+      }
+      throw new InvokeError(
+        InvokeErrorType.UNKNOWN,
+        `HTTP ${response.status}: ${errorMessage}`,
+        errorData
+      );
+    }
+    const data = await response.json();
+    const tool2 = tools2.AgentOutput;
+    const macroToolInput = lenientParseMacroToolCall(data, tool2.inputSchema);
+    let toolResult;
+    try {
+      toolResult = await tool2.execute(macroToolInput);
+    } catch (e) {
+      throw new InvokeError(
+        InvokeErrorType.TOOL_EXECUTION_ERROR,
+        `Tool execution failed: ${e.message}`,
+        e
+      );
+    }
+    return {
+      toolCall: {
+        // id: toolCall.id,
+        name: "AgentOutput",
+        args: macroToolInput
+      },
+      toolResult,
+      usage: {
+        promptTokens: data.usage?.prompt_tokens ?? 0,
+        completionTokens: data.usage?.completion_tokens ?? 0,
+        totalTokens: data.usage?.total_tokens ?? 0,
+        cachedTokens: data.usage?.prompt_tokens_details?.cached_tokens,
+        reasoningTokens: data.usage?.completion_tokens_details?.reasoning_tokens
+      },
+      rawResponse: data
+    };
+  }
+};
+__name(_OpenAIClient, "OpenAIClient");
+let OpenAIClient = _OpenAIClient;
 const _LLM = class _LLM {
   constructor(config, id) {
     __publicField(this, "config");
     __publicField(this, "id");
-    __privateAdd(this, _openai);
-    __privateAdd(this, _model);
+    __publicField(this, "client");
     __privateAdd(this, _bus);
-    this.config = {
-      baseURL: DEFAULT_BASE_URL,
-      apiKey: DEFAULT_API_KEY,
-      modelName: DEFAULT_MODEL_NAME,
-      maxRetries: LLM_MAX_RETRIES,
-      ...config
-    };
+    this.config = parseLLMConfig(config);
     this.id = id;
     __privateSet(this, _bus, getEventBus(id));
-    __privateSet(this, _openai, createOpenAI({ baseURL: this.config.baseURL, apiKey: this.config.apiKey }));
-    __privateSet(this, _model, __privateGet(this, _openai).chat(this.config.modelName));
+    this.client = new OpenAIClient({
+      model: this.config.model,
+      apiKey: this.config.apiKey,
+      baseURL: this.config.baseURL,
+      temperature: this.config.temperature,
+      maxTokens: this.config.maxTokens
+    });
   }
   /**
    * - call llm api *once*
@@ -1596,81 +1900,10 @@ const _LLM = class _LLM {
    * - return the result of the tool
    */
   async invoke(messages, tools2, abortSignal) {
-    const isClaude = this.config.modelName.slice(0, 8).includes("claude");
-    this.config.modelName.slice(0, 6).includes("qwen");
-    this.config.modelName.slice(0, 5).includes("gpt");
     return await withRetry(
       async () => {
-        const result = await generateText({
-          model: __privateGet(this, _model),
-          messages,
-          tools: tools2,
-          abortSignal,
-          /**
-           * 文档中没有说明，从源码看，@facts
-           * - 只会重试被识别为 retryable 的 API_CALL_ERROR
-           * - 返回无法解析的 json 应该不会重试
-           * - experimental_repairToolCall 只会执行一次，不算作重试
-           * @facts
-           * - 许多 proxy 过的 openAI 兼容接口返回的错误格式并不规范，通常不会被识别为 retryable
-           * @conclusion
-           * - 看起来并不实用，不如完全手工控制粗粒度重试
-           */
-          // maxRetries: this.config.maxRetries,
-          maxRetries: 0,
-          // toolChoice: 'required',
-          // @note incompatible to Claude
-          toolChoice: isClaude ? void 0 : { type: "tool", toolName: MACRO_TOOL_NAME },
-          /**
-           * controlled by main loop. our method only call api once
-           */
-          // stopWhen: [hasToolCall('done'), stepCountIs(100)],
-          stopWhen: [stepCountIs(1)],
-          // stopWhen: [hasToolCall('AgentOutput')],
-          providerOptions: {
-            openai: {
-              // @note this one needs all fields in tool schema must be `required`
-              // strictJsonSchema: true,
-              // This way only at most one tool can be called at a time
-              parallelToolCalls: false,
-              reasoningEffort: "minimal",
-              // @note not working
-              // serviceTier: 'priority',
-              textVerbosity: "low",
-              // @note Optimize OpenAI model caching, should be unique per user, currently has no effect
-              promptCacheKey: "page-agent:" + this.id
-            }
-          }
-          /**
-           * schema 出错时执行一次，不确定是否计入重试
-           * 目前看起来像是会直接抛错，被 withRetry 处理
-           * @note
-           * 如果不提供，则 ai-sdk 会把 tool-error 加入 message 中重新调用一次，
-           * 配合 stepCountIs 或者 hasToolCall 都会导致错误被 silent，toolResults 永远为 0
-           * 遗憾的是，这里没有办法抛错（抛错后回到默认逻辑），只要这里 repair 不好，就会导致 silent error
-           * 更糟糕的是，只要传入了 tools，无论 stopWhen 如何设置，都会被当作 multi-step，
-           * 本质上就和我们 single step 的逻辑冲突
-           * 长远来看必须删掉 ai-sdk，直接用 openAI API 实现
-           */
-          // experimental_repairToolCall: (options): Promise<LanguageModelV2ToolCall | null> => {
-          // 	console.error('hahhah', options)
-          // 	throw options.error
-          // },
-        });
-        console.log(chalk.blue.bold("LLM:invoke finished"), result);
-        const toolError = result.content.find((part) => part.type === "tool-error");
-        if (toolError) throw toolError.error;
-        assert(!result.text, "Model returned text without calling done tool", true);
-        assert(result.toolCalls.length === 1, "Model must call exactly one tool", true);
-        assert(result.toolResults.length === 1, "Step must have exactly one tool result", true);
-        const toolCall = result.toolCalls[0];
-        const toolResult = result.toolResults[0];
-        const usage = result.totalUsage;
-        return {
-          toolCall,
-          toolResult,
-          usage
-        };
+        const result = await this.client.invoke(messages, tools2, abortSignal);
+        return result;
       },
       // retry settings
       {
@@ -1691,8 +1924,6 @@ const _LLM = class _LLM {
     );
   }
 };
-_openai = new WeakMap();
-_model = new WeakMap();
 _bus = new WeakMap();
 __name(_LLM, "LLM");
 let LLM = _LLM;
@@ -1710,6 +1941,7 @@ async function withRetry(fn, settings) {
       console.error(error2);
       settings.onError(error2, retries < settings.maxRetries);
       if (error2?.name === "AbortError") throw error2;
+      if (error2 instanceof InvokeError && !error2.retryable) throw error2;
       lastError = error2;
       retries++;
       await new Promise((resolve) => setTimeout(resolve, 100));
@@ -2049,6 +2281,10 @@ const utils = /* @__PURE__ */ Object.freeze(/* @__PURE__ */ Object.definePropert
   waitFor
 }, Symbol.toStringTag, { value: "Module" }));
 window.utils = utils;
+function tool(options) {
+  return options;
+}
+__name(tool, "tool");
 const tools = /* @__PURE__ */ new Map();
 tools.set(
   "done",
@@ -2058,7 +2294,8 @@ tools.set(
       text: zod.string(),
       success: zod.boolean().default(true)
     }),
-    execute: /* @__PURE__ */ __name(function(input2) {
+    execute: /* @__PURE__ */ __name(async function(input2) {
+      return Promise.resolve("Task completed");
     }, "execute")
   })
 );
@@ -2421,7 +2658,7 @@ const _Panel = class _Panel {
       __privateSet(this, _userAnswerResolver, resolve);
       __privateMethod(this, _Panel_instances, update_fn).call(this, {
         type: "output",
-        displayText: `询问: ${question}`
+        displayText: __privateGet(this, _pageAgent).i18n.t("ui.panel.question", { question })
       });
       if (!__privateGet(this, _isExpanded)) {
         __privateMethod(this, _Panel_instances, expand_fn).call(this);
@@ -2482,7 +2719,7 @@ show_fn = /* @__PURE__ */ __name(function() {
   this.wrapper.style.transform = "translateX(-50%) translateY(0)";
 }, "#show");
 /**
- * 隐藏面板
+ * Hide panel
  */
 hide_fn = /* @__PURE__ */ __name(function() {
   this.wrapper.style.opacity = "0";
@@ -2490,7 +2727,7 @@ hide_fn = /* @__PURE__ */ __name(function() {
   this.wrapper.style.display = "none";
 }, "#hide");
 /**
- * 重置状态
+ * Reset state
  */
 reset_fn = /* @__PURE__ */ __name(function() {
   __privateGet(this, _state).reset();
@@ -2511,39 +2748,39 @@ togglePause_fn = /* @__PURE__ */ __name(function() {
   __privateGet(this, _pageAgent).paused = !__privateGet(this, _pageAgent).paused;
   __privateMethod(this, _Panel_instances, updatePauseButton_fn).call(this);
   if (__privateGet(this, _pageAgent).paused) {
-    __privateGet(this, _statusText).textContent = "暂停中，稍后";
+    __privateGet(this, _statusText).textContent = __privateGet(this, _pageAgent).i18n.t("ui.panel.paused");
     __privateMethod(this, _Panel_instances, updateStatusIndicator_fn).call(this, "thinking");
   } else {
-    __privateGet(this, _statusText).textContent = "继续执行";
+    __privateGet(this, _statusText).textContent = __privateGet(this, _pageAgent).i18n.t("ui.panel.continueExecution");
     __privateMethod(this, _Panel_instances, updateStatusIndicator_fn).call(this, "tool_executing");
   }
 }, "#togglePause");
 /**
- * 更新暂停按钮状态
+ * Update pause button state
  */
 updatePauseButton_fn = /* @__PURE__ */ __name(function() {
   if (__privateGet(this, _pageAgent).paused) {
     __privateGet(this, _pauseButton).textContent = "▶";
-    __privateGet(this, _pauseButton).title = "继续";
+    __privateGet(this, _pauseButton).title = __privateGet(this, _pageAgent).i18n.t("ui.panel.continue");
     __privateGet(this, _pauseButton).classList.add(styles$1.paused);
   } else {
     __privateGet(this, _pauseButton).textContent = "⏸︎";
-    __privateGet(this, _pauseButton).title = "暂停";
+    __privateGet(this, _pauseButton).title = __privateGet(this, _pageAgent).i18n.t("ui.panel.pause");
     __privateGet(this, _pauseButton).classList.remove(styles$1.paused);
   }
 }, "#updatePauseButton");
 /**
- * 终止 Agent
+ * Stop Agent
  */
 stopAgent_fn = /* @__PURE__ */ __name(function() {
   __privateMethod(this, _Panel_instances, update_fn).call(this, {
     type: "error",
-    displayText: "任务已终止"
+    displayText: __privateGet(this, _pageAgent).i18n.t("ui.panel.taskTerminated")
   });
   __privateGet(this, _pageAgent).dispose();
 }, "#stopAgent");
 /**
- * 提交任务
+ * Submit task
  */
 submitTask_fn = /* @__PURE__ */ __name(function() {
   const input2 = __privateGet(this, _taskInput).value.trim();
@@ -2556,12 +2793,12 @@ submitTask_fn = /* @__PURE__ */ __name(function() {
   }
 }, "#submitTask");
 /**
- * 处理用户回答
+ * Handle user answer
  */
 handleUserAnswer_fn = /* @__PURE__ */ __name(function(input2) {
   __privateMethod(this, _Panel_instances, update_fn).call(this, {
     type: "input",
-    displayText: `用户回答: ${input2}`
+    displayText: __privateGet(this, _pageAgent).i18n.t("ui.panel.userAnswer", { input: input2 })
   });
   __privateSet(this, _isWaitingForUserAnswer, false);
   if (__privateGet(this, _userAnswerResolver)) {
@@ -2570,24 +2807,24 @@ handleUserAnswer_fn = /* @__PURE__ */ __name(function(input2) {
   }
 }, "#handleUserAnswer");
 /**
- * 显示输入区域
+ * Show input area
  */
 showInputArea_fn = /* @__PURE__ */ __name(function(placeholder) {
   __privateGet(this, _taskInput).value = "";
-  __privateGet(this, _taskInput).placeholder = placeholder || "输入新任务，详细描述步骤，回车提交";
+  __privateGet(this, _taskInput).placeholder = placeholder || __privateGet(this, _pageAgent).i18n.t("ui.panel.taskInput");
   __privateGet(this, _inputSection).classList.remove(styles$1.hidden);
   setTimeout(() => {
     __privateGet(this, _taskInput).focus();
   }, 100);
 }, "#showInputArea");
 /**
- * 隐藏输入区域
+ * Hide input area
  */
 hideInputArea_fn = /* @__PURE__ */ __name(function() {
   __privateGet(this, _inputSection).classList.add(styles$1.hidden);
 }, "#hideInputArea");
 /**
- * 检查是否应该显示输入区域
+ * Check if input area should be shown
  */
 shouldShowInputArea_fn = /* @__PURE__ */ __name(function() {
   if (__privateGet(this, _isWaitingForUserAnswer)) return true;
@@ -2612,23 +2849,23 @@ createWrapper_fn = /* @__PURE__ */ __name(function() {
     stepNumber: 0,
     timestamp: /* @__PURE__ */ new Date(),
     type: "thinking",
-    displayText: "等待任务开始..."
+    displayText: __privateGet(this, _pageAgent).i18n.t("ui.panel.waitingPlaceholder")
   })}
 				</div>
 			</div>
 			<div class="${styles$1.header}">
 				<div class="${styles$1.statusSection}">
 					<div class="${styles$1.indicator} ${styles$1.thinking}"></div>
-					<div class="${styles$1.statusText}">准备就绪</div>
+					<div class="${styles$1.statusText}">${__privateGet(this, _pageAgent).i18n.t("ui.panel.ready")}</div>
 				</div>
 				<div class="${styles$1.controls}">
-					<button class="${styles$1.controlButton} ${styles$1.expandButton}" title="展开历史">
+					<button class="${styles$1.controlButton} ${styles$1.expandButton}" title="${__privateGet(this, _pageAgent).i18n.t("ui.panel.expand")}">
 						▼
 					</button>
-					<button class="${styles$1.controlButton} ${styles$1.pauseButton}" title="暂停">
+					<button class="${styles$1.controlButton} ${styles$1.pauseButton}" title="${__privateGet(this, _pageAgent).i18n.t("ui.panel.pause")}">
 						⏸︎
 					</button>
-					<button class="${styles$1.controlButton} ${styles$1.stopButton}" title="终止">
+					<button class="${styles$1.controlButton} ${styles$1.stopButton}" title="${__privateGet(this, _pageAgent).i18n.t("ui.panel.stop")}">
 						X
 					</button>
 				</div>
@@ -2716,7 +2953,7 @@ updateStatusIndicator_fn = /* @__PURE__ */ __name(function(type) {
 }, "#updateStatusIndicator");
 updateHistory_fn = /* @__PURE__ */ __name(function() {
   const steps = __privateGet(this, _state).getAllSteps();
-  __privateGet(this, _historySection).innerHTML = steps.slice(-10).map((step) => __privateMethod(this, _Panel_instances, createHistoryItem_fn).call(this, step)).join("");
+  __privateGet(this, _historySection).innerHTML = steps.map((step) => __privateMethod(this, _Panel_instances, createHistoryItem_fn).call(this, step)).join("");
   __privateMethod(this, _Panel_instances, scrollToBottom_fn).call(this);
 }, "#updateHistory");
 scrollToBottom_fn = /* @__PURE__ */ __name(function() {
@@ -2735,7 +2972,9 @@ createHistoryItem_fn = /* @__PURE__ */ __name(function(step) {
   let statusIcon2 = "";
   if (step.type === "completed") {
     if (step.toolName === "done") {
-      const isSuccess = !step.toolResult || !step.toolResult.includes("失败") && !step.toolResult.includes("错误");
+      const failureKeyword = __privateGet(this, _pageAgent).i18n.t("ui.tools.resultFailure");
+      const errorKeyword = __privateGet(this, _pageAgent).i18n.t("ui.tools.resultError");
+      const isSuccess = !step.toolResult || !step.toolResult.includes(failureKeyword) && !step.toolResult.includes(errorKeyword);
       typeClass = isSuccess ? styles$1.doneSuccess : styles$1.doneError;
       statusIcon2 = isSuccess ? "🎉" : "❌";
     } else {
@@ -2759,6 +2998,13 @@ createHistoryItem_fn = /* @__PURE__ */ __name(function(step) {
   } else {
     statusIcon2 = "🧠";
   }
+  const durationText = step.duration ? ` · ${step.duration}ms` : "";
+  const stepLabel = __privateGet(this, _pageAgent).i18n.t("ui.panel.step", {
+    number: step.stepNumber.toString(),
+    time,
+    duration: durationText || ""
+    // Explicitly pass empty string to replace template
+  });
   return `
 			<div class="${styles$1.historyItem} ${typeClass}">
 				<div class="${styles$1.historyContent}">
@@ -2766,8 +3012,7 @@ createHistoryItem_fn = /* @__PURE__ */ __name(function(step) {
 					<span>${step.displayText}</span>
 				</div>
 				<div class="${styles$1.historyMeta}">
-					步骤 ${step.stepNumber} · ${time}
-					${step.duration ? ` · ${step.duration}ms` : ""}
+					${stepLabel}
 				</div>
 			</div>
 		`;
@@ -3028,6 +3273,14 @@ moveCursorToTarget_fn = /* @__PURE__ */ __name(function() {
 }, "#moveCursorToTarget");
 __name(_SimulatorMask, "SimulatorMask");
 let SimulatorMask = _SimulatorMask;
+function assert(condition, message, silent) {
+  if (!condition) {
+    const errorMessage = message ?? "Assertion failed";
+    console.error(chalk.red(`❌ assert: ${errorMessage}`));
+    throw new Error(errorMessage);
+  }
+}
+__name(assert, "assert");
 const _PageAgent = class _PageAgent extends EventTarget {
   constructor(config = {}) {
     super();
@@ -3036,6 +3289,7 @@ const _PageAgent = class _PageAgent extends EventTarget {
     __publicField(this, "id", uid());
     __publicField(this, "bus", getEventBus(this.id));
     __publicField(this, "i18n");
+    __publicField(this, "panel");
     __publicField(this, "paused", false);
     __publicField(this, "disposed", false);
     __publicField(this, "task", "");
@@ -3059,13 +3313,12 @@ const _PageAgent = class _PageAgent extends EventTarget {
     __publicField(this, "tools", new Map(tools));
     /** Fullscreen mask */
     __publicField(this, "mask", new SimulatorMask());
-    /** Interactive panel */
-    __publicField(this, "panel", new Panel(this));
     /** History records */
     __publicField(this, "history", []);
     this.config = config;
     __privateSet(this, _llm, new LLM(this.config, this.id));
     this.i18n = new I18n(this.config.language);
+    this.panel = new Panel(this);
     patchReact();
   }
   /**
@@ -3108,18 +3361,16 @@ const _PageAgent = class _PageAgent extends EventTarget {
               content: __privateMethod(this, _PageAgent_instances, assembleUserPrompt_fn).call(this)
             }
           ],
-          // tools,
-          __privateMethod(this, _PageAgent_instances, packMacroTool_fn).call(this),
+          { AgentOutput: __privateMethod(this, _PageAgent_instances, packMacroTool_fn).call(this) },
           __privateGet(this, _abortController).signal
         );
-        const toolResult = result.toolResult;
-        const input2 = toolResult.input;
-        const output2 = toolResult.output;
+        const macroResult = result.toolResult;
+        const input2 = macroResult.input;
+        const output2 = macroResult.output;
         const brain = {
-          thinking: input2.thinking,
-          evaluation_previous_goal: input2.evaluation_previous_goal,
-          memory: input2.memory,
-          next_goal: input2.next_goal
+          evaluation_previous_goal: input2.evaluation_previous_goal || "",
+          memory: input2.memory || "",
+          next_goal: input2.next_goal || ""
         };
         const actionName = Object.keys(input2.action)[0];
         const action = {
@@ -3144,8 +3395,8 @@ const _PageAgent = class _PageAgent extends EventTarget {
           };
         }
         if (actionName === "done") {
-          const success = action.input.success || false;
-          const text = action.input.text || "no text provided";
+          const success = action.input?.success ?? false;
+          const text = action.input?.text || "no text provided";
           console.log(chalk.green.bold("Task completed"), success, text);
           __privateMethod(this, _PageAgent_instances, onDone_fn).call(this, text, success);
           return {
@@ -3190,42 +3441,6 @@ _PageAgent_instances = new WeakSet();
  * - next_goal: string
  * - action: { toolName: toolInput }
  * where action must be selected from tools defined in this.tools
- *
- * @topic 要不要合并成一个 tool？
- * @facts
- * - 我们需要模型每步返回 evaluation/memory/goal 等思考过程
- * - browser use 合并成一个巨大的 tool
- * ```json
- * {
- *  "memory": "...",
- *  "goal": "...",
- *  "actions": [
- *    {
- *      "name": "...",
- *      "args": "..."
- *    }
- *    // ...
- *  ]
- * }
- * ```
- * - qwen 目前必须指定 function name 来确保 tool call
- * @reasoning
- * - 不能为了 qwen 的缺陷而设计系统
- * - 更复杂的 tool 更容易出错
- * - 分散的 tool 更容易利用 ai-sdk 的重试机制，也更容易处理错误
- * - 不能用额外的步骤生成这些数据，不仅性能过差，而且 goal 之类的必须和 call 一起生成
- * @options
- * - Plan @A
- *   - 和 browser use 使用完全一致的做法，合并成一个大 tool，要求每次调用
- *   - 会把 tool 定义变得非常复杂，增加出错率
- * - Plan @B
- *   - 每次调用两个 tool，其中一个用来输出思考
- *   - 很难用提示词 enforce 这么复杂的规则
- * - Plan @C
- *   - 自动为每个 tool 增加固定的 reasoning/memory/goal 等输入，并自动拦截提取这些数据
- *   - 会让 tool 定义变得很长
- * @conclusion
- * - 使用 @A
  */
 packMacroTool_fn = /* @__PURE__ */ __name(function() {
   const tools2 = this.tools;
@@ -3234,70 +3449,75 @@ packMacroTool_fn = /* @__PURE__ */ __name(function() {
       [toolName]: tool2.inputSchema
     });
   });
-  const actionSchema = zod.union(actionSchemas);
+  const actionSchema = zod.union(
+    actionSchemas
+  );
+  const macroToolSchema = zod.object({
+    // thinking: zod.string().optional(),
+    evaluation_previous_goal: zod.string().optional(),
+    memory: zod.string().optional(),
+    next_goal: zod.string().optional(),
+    action: actionSchema
+  });
   return {
-    [MACRO_TOOL_NAME]: tool({
-      // description: 'Output the result of the agent',
-      inputSchema: zod.object({
-        // thinking: zod.string().optional(),
-        evaluation_previous_goal: zod.string().optional(),
-        memory: zod.string().optional(),
-        next_goal: zod.string().optional(),
-        action: actionSchema
-      }),
-      execute: /* @__PURE__ */ __name(async (input2, options) => {
-        if (__privateGet(this, _abortController).signal.aborted) throw new Error("AbortError");
-        await waitUntil(() => !this.paused);
-        console.log(chalk.blue.bold("MacroTool execute"), input2);
-        const action = input2.action;
-        const toolName = Object.keys(action)[0];
-        const toolInput = action[toolName];
-        const brain = trimLines(`✅: ${input2.evaluation_previous_goal}
+    // name: MACRO_TOOL_NAME,
+    // description: 'Execute agent action', // @todo remote
+    inputSchema: macroToolSchema,
+    execute: /* @__PURE__ */ __name(async (input2) => {
+      if (__privateGet(this, _abortController).signal.aborted) throw new Error("AbortError");
+      await waitUntil(() => !this.paused);
+      console.log(chalk.blue.bold("MacroTool execute"), input2);
+      const action = input2.action;
+      const toolName = Object.keys(action)[0];
+      const toolInput = action[toolName];
+      const brain = trimLines(`✅: ${input2.evaluation_previous_goal}
 						💾: ${input2.memory}
 						🎯: ${input2.next_goal}
 					`);
-        console.log(brain);
-        this.bus.emit("panel:update", {
-          type: "thinking",
-          displayText: brain
-        });
-        const tool2 = tools2.get(toolName);
-        assert(tool2, `Tool ${toolName} not found. (@note should have been caught before this!!!)`);
-        console.log(chalk.blue.bold(`Executing tool: ${toolName}`), toolInput, options);
+      console.log(brain);
+      this.bus.emit("panel:update", {
+        type: "thinking",
+        displayText: brain
+      });
+      const tool2 = tools2.get(toolName);
+      assert(tool2, `Tool ${toolName} not found. (@note should have been caught before this!!!)`);
+      console.log(chalk.blue.bold(`Executing tool: ${toolName}`), toolInput);
+      this.bus.emit("panel:update", {
+        type: "tool_executing",
+        toolName,
+        toolArgs: toolInput,
+        displayText: getToolExecutingText(toolName, toolInput, this.i18n)
+      });
+      const startTime = Date.now();
+      let result = await tool2.execute.bind(this)(toolInput);
+      const duration = Date.now() - startTime;
+      console.log(chalk.green.bold(`Tool (${toolName}) executed for ${duration}ms`), result);
+      if (toolName === "wait") {
+        __privateSet(this, _totalWaitTime, __privateGet(this, _totalWaitTime) + Math.round(toolInput.seconds + duration / 1e3));
+        result += `
+<sys> You have waited ${__privateGet(this, _totalWaitTime)} seconds accumulatively.`;
+        if (__privateGet(this, _totalWaitTime) >= 3)
+          result += "\nDo NOT wait any longer unless you have a good reason.\n";
+        result += "</sys>";
+      } else {
+        __privateSet(this, _totalWaitTime, 0);
+      }
+      const displayResult = getToolCompletedText(toolName, toolInput, this.i18n);
+      if (displayResult)
         this.bus.emit("panel:update", {
           type: "tool_executing",
           toolName,
           toolArgs: toolInput,
-          displayText: getToolExecutingText(toolName, toolInput, this.i18n)
+          toolResult: result,
+          displayText: displayResult,
+          duration
         });
-        const startTime = Date.now();
-        let result = await tool2.execute.bind(this)(toolInput, options);
-        const duration = Date.now() - startTime;
-        console.log(chalk.green.bold(`Tool (${toolName}) executed for ${duration}ms`), result);
-        if (toolName === "wait") {
-          __privateSet(this, _totalWaitTime, __privateGet(this, _totalWaitTime) + Math.round(toolInput.seconds + duration / 1e3));
-          result += `
-<sys> You have waited ${__privateGet(this, _totalWaitTime)} seconds accumulatively.`;
-          if (__privateGet(this, _totalWaitTime) >= 3)
-            result += "\nDo NOT wait any longer unless you have a good reason.\n";
-          result += "</sys>";
-        } else {
-          __privateSet(this, _totalWaitTime, 0);
-        }
-        const displayResult = getToolCompletedText(toolName, toolInput, this.i18n);
-        if (displayResult)
-          this.bus.emit("panel:update", {
-            type: "tool_executing",
-            toolName,
-            toolArgs: toolInput,
-            toolResult: result,
-            displayText: displayResult,
-            duration
-          });
-        await new Promise((resolve) => setTimeout(resolve, 100));
-        return result;
-      }, "execute")
-    })
+      await new Promise((resolve) => setTimeout(resolve, 100));
+      return {
+        input: input2,
+        output: result
+      };
+    }, "execute")
   };
 }, "#packMacroTool");
 /**