@eko-ai/eko 1.0.5 → 1.0.6

package/README.md

@@ -1,23 +1,26 @@
 # Eko
 
-[![License](https://img.shields.io/badge/license-MIT-blue.svg)](LICENSE) [![Build Status](https://img.shields.io/badge/build-passing-brightgreen.svg)](https://example.com/build-status) [![Version](https://img.shields.io/badge/version-0.1.0-yellow.svg)](https://example.com/version)
+[![License](https://img.shields.io/badge/license-MIT-blue.svg)](LICENSE) [![Build Status](https://img.shields.io/badge/build-passing-brightgreen.svg)](https://example.com/build-status) [![Version](https://img.shields.io/badge/version-1.0.5-yellow.svg)](https://eko.fellou.ai/docs/release/versions/)
 
 **Eko** is a revolutionary framework designed to empower developers and users alike to program their browser and operating system using natural language. With seamless integration of browser APIs, OS-level capabilities, and cutting-edge AI tools like Claude 3.5, Eko redefines how we interact with technology, making it intuitive, powerful, and accessible.
 
 ## Key Features
 
-- **Natural Language Programming**: Transform human instructions into executable actions using advanced AI models
-- **Cross-Platform Operation**: Run in browser extensions, web applications, or Node.js environments
-- **Powerful Tooling**: Built-in tools for browser automation, OS operations, and system control
-- **Flexible Integration**: Seamless integration with Claude 3.5 and other LLM models
-- **Developer-Friendly**: Comprehensive TypeScript support and extensive documentation
+- **Natural Language Programming**: Transform human instructions into e- **Natural Language Programming**: Convert natural language task descriptions into executable workflows, making agent development more intuitive
+- **Two-Layer Execution Model**: Separate offline planning from online execution, making agent decisions more structured and explainable
+- **Comprehensive Tooling**: Rich built-in tools for browser automation, computer control, file operations, and web interactions
+- **Hybrid Drive System:** Combine LLM capabilities with developer control, enabling "human in the loop" and allowing interference at multiple levels of granularity
+- **Event-Driven Automation:** Trigger workflows based on browser or system events
+- **Environment Flexibility**: Work across different environments ( [Browser Extension](/docs/browseruse/browser-extension), [Web](/docs/browseruse/browser-web), [Node.js](/docs/computeruse/computer-node), [Next-Gen AI Browser Fellou](/docs/computeruse/computer-fellou) ) with consistent APIs
 
-## Quick Start
+## Quickstart
 
 ```bash
 npm install @eko-ai/eko
 ```
 
+> The following code is for reference only. For detailed usage, please refer to the [Eko Quickstart guide](https://eko.fellou.ai/docs/getting-started/quickstart/).
+
 ```typescript
 import { Eko } from '@eko-ai/eko';
 
@@ -26,10 +29,13 @@ const eko = new Eko({
 });
 
 // Example: Browser automation
-await eko.run("Search for 'Eko framework' on Google and save the first result");
+const extWorkflow = await eko.generate("Search for 'Eko framework' on Google and save the first result");
+await eko.execute(extWorkflow);
 
 // Example: System operation
-await eko.run("Create a new folder named 'reports' and move all PDF files there");
+const sysWorkflow = await eko.generate("Create a new folder named 'reports' and move all PDF files there");
+await eko.execute(sysWorkflow);
+
 ```
 
 ## Use Cases

package/dist/extension.cjs.js

@@ -455,7 +455,6 @@ class BrowserUse {
   - 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.
-* \`clear_text\`: Clear the text in the input/textarea element.
 * \`click\`: Click to element.
 * \`right_click\`: Right-click on the element.
 * \`double_click\`: Double-click on the element.
@@ -466,7 +465,6 @@ class BrowserUse {
                     enum: [
                         'screenshot_extract_element',
                         'input_text',
-                        'clear_text',
                         'click',
                         'right_click',
                         'double_click',
@@ -519,16 +517,10 @@ class BrowserUse {
                     if (params.text == null) {
                         throw new Error('text parameter is required');
                     }
+                    await clear_input_by(tabId, selector_xpath, params.index);
                     result = await type_by(tabId, params.text, selector_xpath, params.index);
                     await sleep(200);
                     break;
-                case 'clear_text':
-                    if (params.index == null) {
-                        throw new Error('index parameter is required');
-                    }
-                    result = await clear_input_by(tabId, selector_xpath, params.index);
-                    await sleep(100);
-                    break;
                 case 'click':
                     if (params.index == null) {
                         throw new Error('index parameter is required');

package/dist/extension.esm.js

@@ -453,7 +453,6 @@ class BrowserUse {
   - 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.
-* \`clear_text\`: Clear the text in the input/textarea element.
 * \`click\`: Click to element.
 * \`right_click\`: Right-click on the element.
 * \`double_click\`: Double-click on the element.
@@ -464,7 +463,6 @@ class BrowserUse {
                     enum: [
                         'screenshot_extract_element',
                         'input_text',
-                        'clear_text',
                         'click',
                         'right_click',
                         'double_click',
@@ -517,16 +515,10 @@ class BrowserUse {
                     if (params.text == null) {
                         throw new Error('text parameter is required');
                     }
+                    await clear_input_by(tabId, selector_xpath, params.index);
                     result = await type_by(tabId, params.text, selector_xpath, params.index);
                     await sleep(200);
                     break;
-                case 'clear_text':
-                    if (params.index == null) {
-                        throw new Error('index parameter is required');
-                    }
-                    result = await clear_input_by(tabId, selector_xpath, params.index);
-                    await sleep(100);
-                    break;
                 case 'click':
                     if (params.index == null) {
                         throw new Error('index parameter is required');

package/dist/extension_content_script.js

@@ -123,6 +123,9 @@ function type(request) {
     }
     input.focus && input.focus();
     if (!text) {
+        if (input.value == '') {
+            return true;
+        }
         input.value = '';
     }
     else {

package/dist/index.cjs.js

@@ -16,11 +16,15 @@ class WorkflowImpl {
         if (!this.validateDAG()) {
             throw new Error("Invalid workflow: Contains circular dependencies");
         }
+        this.abort = false;
         callback && await ((_b = (_a = callback.hooks).beforeWorkflow) === null || _b === void 0 ? void 0 : _b.call(_a, this));
         const executed = new Set();
         const executing = new Set();
         const executeNode = async (nodeId) => {
             var _a, _b, _c, _d, _e;
+            if (this.abort) {
+                throw new Error("Abort");
+            }
             if (executed.has(nodeId)) {
                 return;
             }
@@ -30,12 +34,22 @@ class WorkflowImpl {
             const node = this.getNode(nodeId);
             // Execute the node's action
             const context = {
+                __skip: false,
+                __abort: false,
                 variables: this.variables,
                 llmProvider: this.llmProvider,
                 tools: new Map(node.action.tools.map(tool => [tool.name, tool])),
-                callback
+                callback,
+                next: () => context.__skip = true,
+                abortAll: () => this.abort = context.__abort = true,
             };
             callback && await ((_b = (_a = callback.hooks).beforeSubtask) === null || _b === void 0 ? void 0 : _b.call(_a, node, context));
+            if (context.__abort) {
+                throw new Error("Abort");
+            }
+            else if (context.__skip) {
+                return;
+            }
             executing.add(nodeId);
             // Execute dependencies first
             for (const depId of node.dependencies) {
@@ -221,12 +235,16 @@ class ActionImpl {
                 toolExecutionPromise = (async () => {
                     try {
                         // beforeToolUse
+                        context.__skip = false;
                         if (context.callback && context.callback.hooks.beforeToolUse) {
                             let modified_input = await context.callback.hooks.beforeToolUse(tool, context, toolCall.input);
                             if (modified_input) {
                                 toolCall.input = modified_input;
                             }
                         }
+                        if (context.__skip || context.__abort) {
+                            return;
+                        }
                         // Execute the tool
                         let result = await tool.execute(context, toolCall.input);
                         // afterToolUse
@@ -291,6 +309,9 @@ class ActionImpl {
         if (toolExecutionPromise) {
             await toolExecutionPromise;
         }
+        if (context.__abort) {
+            throw new Error("Abort");
+        }
         // Add messages in the correct order after everything is complete
         if (assistantTextMessage) {
             roundMessages.push({ role: 'assistant', content: assistantTextMessage });
@@ -440,7 +461,7 @@ ${toolDescriptions}
 Generate a complete workflow that:
 1. Only uses the tools listed above
 2. Properly sequences tool usage based on dependencies
-3. Ensures each action has appropriate input/output schemas
+3. Ensures each action has appropriate input/output schemas, and that the "tools" field in each action is populated with the sufficient subset of all available tools needed to complete the action
 4. Creates a clear, logical flow to accomplish the user's goal
 5. Includes detailed descriptions for each action, ensuring that the actions, when combined, is a complete solution to the user's problem`;
         },

package/dist/index.esm.js

@@ -12,11 +12,15 @@ class WorkflowImpl {
         if (!this.validateDAG()) {
             throw new Error("Invalid workflow: Contains circular dependencies");
         }
+        this.abort = false;
         callback && await ((_b = (_a = callback.hooks).beforeWorkflow) === null || _b === void 0 ? void 0 : _b.call(_a, this));
         const executed = new Set();
         const executing = new Set();
         const executeNode = async (nodeId) => {
             var _a, _b, _c, _d, _e;
+            if (this.abort) {
+                throw new Error("Abort");
+            }
             if (executed.has(nodeId)) {
                 return;
             }
@@ -26,12 +30,22 @@ class WorkflowImpl {
             const node = this.getNode(nodeId);
             // Execute the node's action
             const context = {
+                __skip: false,
+                __abort: false,
                 variables: this.variables,
                 llmProvider: this.llmProvider,
                 tools: new Map(node.action.tools.map(tool => [tool.name, tool])),
-                callback
+                callback,
+                next: () => context.__skip = true,
+                abortAll: () => this.abort = context.__abort = true,
             };
             callback && await ((_b = (_a = callback.hooks).beforeSubtask) === null || _b === void 0 ? void 0 : _b.call(_a, node, context));
+            if (context.__abort) {
+                throw new Error("Abort");
+            }
+            else if (context.__skip) {
+                return;
+            }
             executing.add(nodeId);
             // Execute dependencies first
             for (const depId of node.dependencies) {
@@ -217,12 +231,16 @@ class ActionImpl {
                 toolExecutionPromise = (async () => {
                     try {
                         // beforeToolUse
+                        context.__skip = false;
                         if (context.callback && context.callback.hooks.beforeToolUse) {
                             let modified_input = await context.callback.hooks.beforeToolUse(tool, context, toolCall.input);
                             if (modified_input) {
                                 toolCall.input = modified_input;
                             }
                         }
+                        if (context.__skip || context.__abort) {
+                            return;
+                        }
                         // Execute the tool
                         let result = await tool.execute(context, toolCall.input);
                         // afterToolUse
@@ -287,6 +305,9 @@ class ActionImpl {
         if (toolExecutionPromise) {
             await toolExecutionPromise;
         }
+        if (context.__abort) {
+            throw new Error("Abort");
+        }
         // Add messages in the correct order after everything is complete
         if (assistantTextMessage) {
             roundMessages.push({ role: 'assistant', content: assistantTextMessage });
@@ -436,7 +457,7 @@ ${toolDescriptions}
 Generate a complete workflow that:
 1. Only uses the tools listed above
 2. Properly sequences tool usage based on dependencies
-3. Ensures each action has appropriate input/output schemas
+3. Ensures each action has appropriate input/output schemas, and that the "tools" field in each action is populated with the sufficient subset of all available tools needed to complete the action
 4. Creates a clear, logical flow to accomplish the user's goal
 5. Includes detailed descriptions for each action, ensuring that the actions, when combined, is a complete solution to the user's problem`;
         },

package/dist/models/workflow.d.ts

@@ -6,6 +6,7 @@ export declare class WorkflowImpl implements Workflow {
     nodes: WorkflowNode[];
     variables: Map<string, unknown>;
     llmProvider?: LLMProvider | undefined;
+    abort?: boolean;
     constructor(id: string, name: string, description?: string | undefined, nodes?: WorkflowNode[], variables?: Map<string, unknown>, llmProvider?: LLMProvider | undefined);
     execute(callback?: WorkflowCallback): Promise<void>;
     addNode(node: WorkflowNode): void;

package/dist/web.cjs.js

@@ -8581,6 +8581,9 @@ function do_input(text, xpath, highlightIndex) {
     }
     input.focus && input.focus();
     if (!text) {
+        if (input.value == '') {
+            return true;
+        }
         input.value = '';
     }
     else {
@@ -8682,7 +8685,6 @@ class BrowserUse {
   - 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.
-* \`clear_text\`: Clear the text in the input/textarea element.
 * \`click\`: Click to element.
 * \`right_click\`: Right-click on the element.
 * \`double_click\`: Double-click on the element.
@@ -8693,7 +8695,6 @@ class BrowserUse {
                     enum: [
                         'screenshot_extract_element',
                         'input_text',
-                        'clear_text',
                         'click',
                         'right_click',
                         'double_click',
@@ -8744,16 +8745,10 @@ class BrowserUse {
                     if (params.text == null) {
                         throw new Error('text parameter is required');
                     }
+                    await clear_input(selector_xpath, params.index);
                     result = await type(params.text, selector_xpath, params.index);
                     await sleep(200);
                     break;
-                case 'clear_text':
-                    if (params.index == null) {
-                        throw new Error('index parameter is required');
-                    }
-                    result = await clear_input(selector_xpath, params.index);
-                    await sleep(100);
-                    break;
                 case 'click':
                     if (params.index == null) {
                         throw new Error('index parameter is required');

package/dist/web.esm.js

@@ -8579,6 +8579,9 @@ function do_input(text, xpath, highlightIndex) {
     }
     input.focus && input.focus();
     if (!text) {
+        if (input.value == '') {
+            return true;
+        }
         input.value = '';
     }
     else {
@@ -8680,7 +8683,6 @@ class BrowserUse {
   - 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.
-* \`clear_text\`: Clear the text in the input/textarea element.
 * \`click\`: Click to element.
 * \`right_click\`: Right-click on the element.
 * \`double_click\`: Double-click on the element.
@@ -8691,7 +8693,6 @@ class BrowserUse {
                     enum: [
                         'screenshot_extract_element',
                         'input_text',
-                        'clear_text',
                         'click',
                         'right_click',
                         'double_click',
@@ -8742,16 +8743,10 @@ class BrowserUse {
                     if (params.text == null) {
                         throw new Error('text parameter is required');
                     }
+                    await clear_input(selector_xpath, params.index);
                     result = await type(params.text, selector_xpath, params.index);
                     await sleep(200);
                     break;
-                case 'clear_text':
-                    if (params.index == null) {
-                        throw new Error('index parameter is required');
-                    }
-                    result = await clear_input(selector_xpath, params.index);
-                    await sleep(100);
-                    break;
                 case 'click':
                     if (params.index == null) {
                         throw new Error('index parameter is required');

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@eko-ai/eko",
-  "version": "1.0.5",
+  "version": "1.0.6",
   "description": "Empowering language to transform human words into action.",
   "main": "dist/index.cjs.js",
   "module": "dist/index.esm.js",