@eko-ai/eko 1.0.9 → 1.0.11

package/dist/extension/script/common.js

@@ -88,10 +88,14 @@ eko.click = function(element) {
 eko.sendKeys = function(element, str, clear, keypress) {
     element.focus && element.focus()
     if (clear) {
-        for (let i = 0; i < element.value.length; i++) {
-            element.dispatchEvent(new KeyboardEvent('keydown', { key: 'Backspace' }))
+        if (element.value == undefined) {
+            element.textContent = ''
+        } else {
+            for (let i = 0; i < element.value.length; i++) {
+                element.dispatchEvent(new KeyboardEvent('keydown', { key: 'Backspace' }))
+            }
+            element.value = ''
         }
-        element.value = ''
     }
     if (keypress) {
         Array.from(str).forEach(key => {
@@ -100,7 +104,11 @@ eko.sendKeys = function(element, str, clear, keypress) {
         element.value += str
         element.dispatchEvent(new Event('input'))
     } else {
-        element.value += str
+        if (element.value == undefined) {
+          element.textContent += str
+        } else {
+          element.value += str
+        }
         element.dispatchEvent(new Event('input', { bubbles: true }))
     }
 }

package/dist/extension/tools/cancel_workflow.d.ts

@@ -0,0 +1,9 @@
+import { CancelWorkflowInput } from '../../types/tools.types';
+import { Tool, InputSchema, ExecutionContext } from '../../types/action.types';
+export declare class CancelWorkflow implements Tool<CancelWorkflowInput, void> {
+    name: string;
+    description: string;
+    input_schema: InputSchema;
+    constructor();
+    execute(context: ExecutionContext, params: CancelWorkflowInput): Promise<void>;
+}

package/dist/extension/utils.d.ts

@@ -1,7 +1,7 @@
 import { ExecutionContext } from '../types/action.types';
 export declare function getWindowId(context: ExecutionContext): Promise<number>;
 export declare function getTabId(context: ExecutionContext): Promise<number>;
-export declare function getCurrentTabId(windowId?: number): Promise<number>;
+export declare function getCurrentTabId(windowId?: number | undefined): Promise<number | undefined>;
 export declare function open_new_tab(url: string, newWindow: boolean, windowId?: number): Promise<chrome.tabs.Tab>;
 export declare function executeScript(tabId: number, func: any, args: any[]): Promise<any>;
 export declare function waitForTabComplete(tabId: number, timeout?: number): Promise<chrome.tabs.Tab>;

package/dist/extension.cjs.js

@@ -30,21 +30,12 @@ async function getWindowId(context) {
 async function getTabId(context) {
     let tabId = context.variables.get('tabId');
     if (tabId) {
-        // Ensure tabId is a number
-        tabId = Number(tabId);
-        // Check if it's a valid integer
-        if (!Number.isInteger(tabId)) {
-            context.variables.delete('tabId');
-            tabId = null;
+        try {
+            await chrome.tabs.get(tabId);
         }
-        else {
-            try {
-                await chrome.tabs.get(tabId);
-            }
-            catch (e) {
-                tabId = null;
-                context.variables.delete('tabId');
-            }
+        catch (e) {
+            tabId = null;
+            context.variables.delete('tabId');
         }
     }
     if (!tabId) {
@@ -68,24 +59,30 @@ async function getTabId(context) {
     }
     return tabId;
 }
-async function getCurrentTabId(windowId) {
+function getCurrentTabId(windowId) {
     return new Promise((resolve, reject) => {
-        chrome.tabs.query({ active: true, currentWindow: true }, function (tabs) {
+        chrome.tabs.query({ windowId, active: true, lastFocusedWindow: true }, function (tabs) {
             if (chrome.runtime.lastError) {
                 console.error('Chrome runtime error:', chrome.runtime.lastError);
                 reject(chrome.runtime.lastError);
                 return;
             }
-            if (!tabs || tabs.length === 0) {
-                reject(new Error('No active tab found'));
-                return;
+            if (tabs.length > 0) {
+                resolve(tabs[0].id);
             }
-            const tabId = tabs[0].id;
-            if (typeof tabId !== 'number') {
-                reject(new Error('Invalid tab ID'));
-                return;
+            else {
+                chrome.tabs.query({ windowId, active: true, currentWindow: true }, function (_tabs) {
+                    if (_tabs.length > 0) {
+                        resolve(_tabs[0].id);
+                        return;
+                    }
+                    else {
+                        chrome.tabs.query({ windowId, status: 'complete', currentWindow: true }, function (__tabs) {
+                            resolve(__tabs.length ? __tabs[__tabs.length - 1].id : undefined);
+                        });
+                    }
+                });
             }
-            resolve(tabId);
         });
     });
 }
@@ -647,35 +644,35 @@ var browser = /*#__PURE__*/Object.freeze({
 class BrowserUse {
     constructor() {
         this.name = 'browser_use';
-        this.description = `Use structured commands to interact with the browser, manipulating page elements through screenshots and webpage element extraction.
-* This is a browser GUI interface where you need to analyze webpages by taking screenshots and extracting page element structures, and specify action sequences to complete designated tasks.
-* Before any operation, you must first call the \`screenshot_extract_element\` command, which will return the browser page screenshot and structured element information, both specially processed.
-* ELEMENT INTERACTION:
-   - Only use indexes that exist in the provided element list
-   - Each element has a unique index number (e.g., "[33]:<button>")
-   - Elements marked with "[]:" are non-interactive (for context only)
-* NAVIGATION & ERROR HANDLING:
-   - If no suitable elements exist, use other functions to complete the task
-   - If stuck, try alternative approaches
-   - Handle popups/cookies by accepting or closing them
+        this.description = `Use structured commands to interact with the browser, manipulating page elements through screenshots and webpage element extraction.
+* This is a browser GUI interface where you need to analyze webpages by taking screenshots and extracting page element structures, and specify action sequences to complete designated tasks.
+* Before any operation, you must first call the \`screenshot_extract_element\` command, which will return the browser page screenshot and structured element information, both specially processed.
+* ELEMENT INTERACTION:
+   - Only use indexes that exist in the provided element list
+   - Each element has a unique index number (e.g., "[33]:<button>")
+   - Elements marked with "[]:" are non-interactive (for context only)
+* NAVIGATION & ERROR HANDLING:
+   - If no suitable elements exist, use other functions to complete the task
+   - If stuck, try alternative approaches
+   - Handle popups/cookies by accepting or closing them
    - Use scroll to find elements you are looking for`;
         this.input_schema = {
             type: 'object',
             properties: {
                 action: {
                     type: 'string',
-                    description: `The action to perform. The available actions are:
-* \`screenshot_extract_element\`: Take a screenshot of the web page and extract operable elements.
-  - Screenshots are used to understand page layouts, with labeled bounding boxes corresponding to element indexes. Each bounding box and its label share the same color, with labels typically positioned in the top-right corner of the box.
-  - Screenshots help verify element positions and relationships. Labels may sometimes overlap, so extracted elements are used to verify the correct elements.
-  - In addition to screenshots, simplified information about interactive elements is returned, with element indexes corresponding to those in the screenshots.
-* \`input_text\`: Enter a string in the interactive element.
-* \`click\`: Click to element.
-* \`right_click\`: Right-click on the element.
-* \`double_click\`: Double-click on the element.
-* \`scroll_to\`: Scroll to the specified element.
-* \`extract_content\`: Extract the text content of the current webpage.
-* \`get_dropdown_options\`: Get all options from a native dropdown element.
+                    description: `The action to perform. The available actions are:
+* \`screenshot_extract_element\`: Take a screenshot of the web page and extract operable elements.
+  - Screenshots are used to understand page layouts, with labeled bounding boxes corresponding to element indexes. Each bounding box and its label share the same color, with labels typically positioned in the top-right corner of the box.
+  - Screenshots help verify element positions and relationships. Labels may sometimes overlap, so extracted elements are used to verify the correct elements.
+  - In addition to screenshots, simplified information about interactive elements is returned, with element indexes corresponding to those in the screenshots.
+* \`input_text\`: Enter a string in the interactive element.
+* \`click\`: Click to element.
+* \`right_click\`: Right-click on the element.
+* \`double_click\`: Double-click on the element.
+* \`scroll_to\`: Scroll to the specified element.
+* \`extract_content\`: Extract the text content of the current webpage.
+* \`get_dropdown_options\`: Get all options from a native dropdown element.
 * \`select_dropdown_option\`: Select dropdown option for interactive element index by the text of the option you want to select.`,
                     enum: [
                         'screenshot_extract_element',
@@ -828,7 +825,7 @@ class BrowserUse {
         }
         catch (e) {
             console.error('Browser use error:', e);
-            return { success: false, error: e === null || e === void 0 ? void 0 : e.message };
+            return { success: false, error: e === null || e === undefined ? undefined : e.message };
         }
     }
     destroy(context) {
@@ -906,7 +903,7 @@ function extractOperableElements() {
                 return NodeFilter.FILTER_SKIP;
             }
             // text <= 100
-            const text = (_a = node.innerText) === null || _a === void 0 ? void 0 : _a.trim();
+            const text = (_a = node.innerText) === null || _a === undefined ? undefined : _a.trim();
             if (isElementVisible(node) &&
                 text &&
                 text.length <= 100 &&
@@ -1012,21 +1009,21 @@ async function executeWithHtmlElement$1(context, task_prompt) {
     let messages = [
         {
             role: 'user',
-            content: `# Task
-Determine the operation intent based on user input, find the element ID that the user needs to operate on in the webpage HTML, and if the element does not exist, do nothing.
-Output JSON format, no explanation required.
-
-# User input
-${task_prompt}
-
-# Output example (when the element exists)
-{"elementId": "1", "operationType": "click"}
-
-# Output example (when the element does not exist)
-{"elementId": null, "operationType": "unknown"}
-
-# HTML
-${pseudoHtml}
+            content: `# Task
+Determine the operation intent based on user input, find the element ID that the user needs to operate on in the webpage HTML, and if the element does not exist, do nothing.
+Output JSON format, no explanation required.
+
+# User input
+${task_prompt}
+
+# Output example (when the element exists)
+{"elementId": "1", "operationType": "click"}
+
+# Output example (when the element does not exist)
+{"elementId": null, "operationType": "unknown"}
+
+# HTML
+${pseudoHtml}
 `,
         },
     ];
@@ -1167,8 +1164,15 @@ class ExportFile {
             });
         }
         catch (e) {
-            let tab = await open_new_tab('https://www.google.com', true);
-            (_c = (_b = (_a = context.callback) === null || _a === void 0 ? void 0 : _a.hooks) === null || _b === void 0 ? void 0 : _b.onTabCreated) === null || _c === void 0 ? void 0 : _c.call(_b, tab.id);
+            let tab;
+            const url = 'https://www.google.com';
+            if (context.ekoConfig.workingWindowId) {
+                tab = await open_new_tab(url, false, context.ekoConfig.workingWindowId);
+            }
+            else {
+                tab = await open_new_tab(url, true);
+            }
+            (_c = (_b = (_a = context.callback) === null || _a === undefined ? undefined : _a.hooks) === null || _b === undefined ? undefined : _b.onTabCreated) === null || _c === undefined ? undefined : _c.call(_b, tab.id);
             let tabId = tab.id;
             await chrome.scripting.executeScript({
                 target: { tabId: tabId },
@@ -1262,21 +1266,21 @@ async function executeWithHtmlElement(context, task_prompt) {
     let messages = [
         {
             role: 'user',
-            content: `# Task
-Find the element ID that the user needs to operate on in the webpage HTML, and if the element does not exist, do nothing.
-Output JSON format, no explanation required.
-
-# User input
-${task_prompt}
-
-# Output example (when the element exists)
-{"elementId": "1"}
-
-# Output example (when the element does not exist)
-{"elementId": null}
-
-# HTML
-${pseudoHtml}
+            content: `# Task
+Find the element ID that the user needs to operate on in the webpage HTML, and if the element does not exist, do nothing.
+Output JSON format, no explanation required.
+
+# User input
+${task_prompt}
+
+# Output example (when the element exists)
+{"elementId": "1"}
+
+# Output example (when the element does not exist)
+{"elementId": null}
+
+# HTML
+${pseudoHtml}
 `,
         },
     ];
@@ -1425,12 +1429,12 @@ class OpenUrl {
         let tab;
         if (newWindow) {
             tab = await open_new_tab(url, true);
-            (_c = (_b = (_a = context.callback) === null || _a === void 0 ? void 0 : _a.hooks) === null || _b === void 0 ? void 0 : _b.onTabCreated) === null || _c === void 0 ? void 0 : _c.call(_b, tab.id);
+            (_c = (_b = (_a = context.callback) === null || _a === undefined ? undefined : _a.hooks) === null || _b === undefined ? undefined : _b.onTabCreated) === null || _c === undefined ? undefined : _c.call(_b, tab.id);
         }
         else {
             let windowId = await getWindowId(context);
             tab = await open_new_tab(url, false, windowId);
-            (_f = (_e = (_d = context.callback) === null || _d === void 0 ? void 0 : _d.hooks) === null || _e === void 0 ? void 0 : _e.onTabCreated) === null || _f === void 0 ? void 0 : _f.call(_e, tab.id);
+            (_f = (_e = (_d = context.callback) === null || _d === undefined ? undefined : _d.hooks) === null || _e === undefined ? undefined : _e.onTabCreated) === null || _f === undefined ? undefined : _f.call(_e, tab.id);
         }
         let windowId = tab.windowId;
         let tabId = tab.id;
@@ -1489,13 +1493,13 @@ class TabManagement {
             properties: {
                 command: {
                     type: 'string',
-                    description: `The command to perform. The available commands are:
-* \`tab_all\`: View all tabs and return the tabId and title.
-* \`current_tab\`: Get current tab information (tabId, url, title).
-* \`go_back\`: Go back to the previous page in the current tab.
-* \`change_url [url]\`: open URL in the current tab, eg: \`change_url https://www.google.com\`.
-* \`close_tab\`: Close the current tab.
-* \`switch_tab [tabId]\`: Switch to the specified tab using tabId, eg: \`switch_tab 1000\`.
+                    description: `The command to perform. The available commands are:
+* \`tab_all\`: View all tabs and return the tabId and title.
+* \`current_tab\`: Get current tab information (tabId, url, title).
+* \`go_back\`: Go back to the previous page in the current tab.
+* \`change_url [url]\`: open URL in the current tab, eg: \`change_url https://www.google.com\`.
+* \`close_tab\`: Close the current tab.
+* \`switch_tab [tabId]\`: Switch to the specified tab using tabId, eg: \`switch_tab 1000\`.
 * \`new_tab [url]\`: Open a new tab window and open the URL, eg: \`new_tab https://www.google.com\``,
                 },
             },
@@ -1596,12 +1600,12 @@ class TabManagement {
             let tab;
             if (newWindow) {
                 tab = await open_new_tab(url, true);
-                (_c = (_b = (_a = context.callback) === null || _a === void 0 ? void 0 : _a.hooks) === null || _b === void 0 ? void 0 : _b.onTabCreated) === null || _c === void 0 ? void 0 : _c.call(_b, tab.id);
+                (_c = (_b = (_a = context.callback) === null || _a === undefined ? undefined : _a.hooks) === null || _b === undefined ? undefined : _b.onTabCreated) === null || _c === undefined ? undefined : _c.call(_b, tab.id);
             }
             else {
                 let windowId = await getWindowId(context);
                 tab = await open_new_tab(url, false, windowId);
-                (_f = (_e = (_d = context.callback) === null || _d === void 0 ? void 0 : _d.hooks) === null || _e === void 0 ? void 0 : _e.onTabCreated) === null || _f === void 0 ? void 0 : _f.call(_e, tab.id);
+                (_f = (_e = (_d = context.callback) === null || _d === undefined ? undefined : _d.hooks) === null || _e === undefined ? undefined : _e.onTabCreated) === null || _f === undefined ? undefined : _f.call(_e, tab.id);
             }
             let windowId = tab.windowId;
             let tabId = tab.id;
@@ -1680,8 +1684,8 @@ class WebSearch {
         }
         let taskId = new Date().getTime() + '';
         let searchs = [{ url: url, keyword: query }];
-        let searchInfo = await deepSearch(context, taskId, searchs, maxResults || 5);
-        let links = ((_a = searchInfo.result[0]) === null || _a === void 0 ? void 0 : _a.links) || [];
+        let searchInfo = await deepSearch(context, taskId, searchs, maxResults || 5, context.ekoConfig.workingWindowId);
+        let links = ((_a = searchInfo.result[0]) === null || _a === undefined ? undefined : _a.links) || [];
         return links.filter((s) => s.content);
     }
 }
@@ -1753,25 +1757,27 @@ chrome.tabs.onUpdated.addListener(async function (tabId, changeInfo, tab) {
  * @param {array} searchs search list => [{ url: 'https://bing.com', keyword: 'ai' }]
  * @param {number} detailsMaxNum Maximum crawling quantity per search detail page
  */
-async function deepSearch(context, taskId, searchs, detailsMaxNum, window) {
+async function deepSearch(context, taskId, searchs, detailsMaxNum, windowId) {
     let closeWindow = false;
-    if (!window) {
+    if (!windowId) {
         // open new window
-        window = await chrome.windows.create({
+        let window = await chrome.windows.create({
             type: 'normal',
             state: 'maximized',
             url: null,
         });
+        windowId = window.id;
         closeWindow = true;
     }
+    windowId = windowId;
     // crawler the search page details page link
     // [{ links: [{ title, url }] }]
-    let detailLinkGroups = await doDetailLinkGroups(context, taskId, searchs, detailsMaxNum, window);
+    let detailLinkGroups = await doDetailLinkGroups(context, taskId, searchs, detailsMaxNum, windowId);
     // crawler all details page content and comments
-    let searchInfo = await doPageContent(context, taskId, detailLinkGroups, window);
+    let searchInfo = await doPageContent(context, taskId, detailLinkGroups, windowId);
     console.log('searchInfo: ', searchInfo);
     // close window
-    closeWindow && chrome.windows.remove(window.id);
+    closeWindow && chrome.windows.remove(windowId);
     return searchInfo;
 }
 /**
@@ -1783,7 +1789,7 @@ async function deepSearch(context, taskId, searchs, detailsMaxNum, window) {
  * @param {*} window
  * @returns [{ links: [{ title, url }] }]
  */
-async function doDetailLinkGroups(context, taskId, searchs, detailsMaxNum, window) {
+async function doDetailLinkGroups(context, taskId, searchs, detailsMaxNum, windowId) {
     var _a, _b, _c;
     let detailLinkGroups = [];
     let countDownLatch = new CountDownLatch(searchs.length);
@@ -1794,7 +1800,7 @@ async function doDetailLinkGroups(context, taskId, searchs, detailsMaxNum, windo
             // open new Tab
             let tab = await chrome.tabs.create({
                 url: url,
-                windowId: window.id,
+                windowId,
             });
             (_c = (_b = (_a = context.callback) === null || _a === void 0 ? void 0 : _a.hooks) === null || _b === void 0 ? void 0 : _b.onTabCreated) === null || _c === void 0 ? void 0 : _c.call(_b, tab.id);
             let eventId = taskId + '_' + i;
@@ -1847,7 +1853,7 @@ async function doDetailLinkGroups(context, taskId, searchs, detailsMaxNum, windo
  * @param {*} window
  * @returns search info
  */
-async function doPageContent(context, taskId, detailLinkGroups, window) {
+async function doPageContent(context, taskId, detailLinkGroups, windowId) {
     var _a, _b, _c;
     const searchInfo = {
         total: 0,
@@ -1870,9 +1876,9 @@ async function doPageContent(context, taskId, detailLinkGroups, window) {
             // open new tab
             let tab = await chrome.tabs.create({
                 url: link.url,
-                windowId: window.id,
+                windowId,
             });
-            (_c = (_b = (_a = context.callback) === null || _a === void 0 ? void 0 : _a.hooks) === null || _b === void 0 ? void 0 : _b.onTabCreated) === null || _c === void 0 ? void 0 : _c.call(_b, tab.id);
+            (_c = (_b = (_a = context.callback) === null || _a === undefined ? undefined : _a.hooks) === null || _b === undefined ? undefined : _b.onTabCreated) === null || _c === undefined ? undefined : _c.call(_b, tab.id);
             searchInfo.running++;
             let eventId = taskId + '_' + i + '_' + j;
             // Create a timeout promise
@@ -2046,7 +2052,7 @@ class CancelWorkflow {
         }
         const reason = params.reason;
         console.log("The workflow has been cancelled because: " + reason);
-        await ((_a = context.workflow) === null || _a === void 0 ? void 0 : _a.cancel());
+        await ((_a = context.workflow) === null || _a === undefined ? undefined : _a.cancel());
         return;
     }
 }
@@ -2073,7 +2079,7 @@ class HumanInputText {
         }
         const question = params.question;
         console.log("question: " + question);
-        let answer = await ((_c = (_a = context.callback) === null || _a === void 0 ? void 0 : (_b = _a.hooks).onHumanInputText) === null || _c === void 0 ? void 0 : _c.call(_b, question));
+        let answer = await ((_c = (_a = context.callback) === null || _a === undefined ? undefined : (_b = _a.hooks).onHumanInputText) === null || _c === undefined ? undefined : _c.call(_b, question));
         if (!answer) {
             console.error("Cannot get user's answer.");
             return { status: "Error: Cannot get user's answer.", answer: "" };
@@ -2112,7 +2118,7 @@ class HumanInputSingleChoice {
         const choices = params.choices;
         console.log("question: " + question);
         console.log("choices: " + choices);
-        let answer = await ((_c = (_a = context.callback) === null || _a === void 0 ? void 0 : (_b = _a.hooks).onHumanInputSingleChoice) === null || _c === void 0 ? void 0 : _c.call(_b, question, choices));
+        let answer = await ((_c = (_a = context.callback) === null || _a === undefined ? undefined : (_b = _a.hooks).onHumanInputSingleChoice) === null || _c === undefined ? undefined : _c.call(_b, question, choices));
         if (!answer) {
             console.error("Cannot get user's answer.");
             return { status: "Error: Cannot get user's answer.", answer: "" };
@@ -2151,7 +2157,7 @@ class HumanInputMultipleChoice {
         const choices = params.choices;
         console.log("question: " + question);
         console.log("choices: " + choices);
-        let answer = await ((_c = (_a = context.callback) === null || _a === void 0 ? void 0 : (_b = _a.hooks).onHumanInputMultipleChoice) === null || _c === void 0 ? void 0 : _c.call(_b, question, choices));
+        let answer = await ((_c = (_a = context.callback) === null || _a === undefined ? undefined : (_b = _a.hooks).onHumanInputMultipleChoice) === null || _c === undefined ? undefined : _c.call(_b, question, choices));
         if (!answer) {
             console.error("Cannot get user's answer.");
             return { status: "Error: Cannot get user's answer.", answer: [] };
@@ -2184,7 +2190,7 @@ class HumanOperate {
         }
         const reason = params.reason;
         console.log("reason: " + reason);
-        let userOperation = await ((_c = (_a = context.callback) === null || _a === void 0 ? void 0 : (_b = _a.hooks).onHumanOperate) === null || _c === void 0 ? void 0 : _c.call(_b, reason));
+        let userOperation = await ((_c = (_a = context.callback) === null || _a === undefined ? undefined : (_b = _a.hooks).onHumanOperate) === null || _c === undefined ? undefined : _c.call(_b, reason));
         if (!userOperation) {
             console.error("Cannot get user's operation.");
             return { status: "Error: Cannot get user's operation.", userOperation: "" };
@@ -2218,7 +2224,7 @@ class SummaryWorkflow {
         }
         const summary = params.summary;
         console.log("summary: " + summary);
-        await ((_c = (_a = context.callback) === null || _a === void 0 ? void 0 : (_b = _a.hooks).onSummaryWorkflow) === null || _c === void 0 ? void 0 : _c.call(_b, summary));
+        await ((_c = (_a = context.callback) === null || _a === undefined ? undefined : (_b = _a.hooks).onSummaryWorkflow) === null || _c === undefined ? undefined : _c.call(_b, summary));
         return { status: "OK" };
     }
 }