-
-
- 
-
-- Introduction • - Demo • - Tutorial • - API• - OpenApi -
+| Skill | Tasks | +| --- | --- | +| [vchart-skill](skills/vchart-skill/SKILL.md) | Chart generation, diagnosis, image replication and React integration | +| [vtable-skill](skills/vtable-skill/SKILL.md) | List tables, pivot tables/charts, interactions and diagnosis | - - -[](https://www.npmjs.com/package/@visactor/vmind) -[](https://www.npmjs.com/package/@visactor/vmind) -[](https://github.com/VisActor/VMind/blob/main/CONTRIBUTING.md#your-first-pull-request) +Each skill includes its instructions, local references, templates and scripts. No separate VisActor CLI, MCP server, model key or hosted VisActor API is required. The host agent supplies the model. - [](https://github.com/visactor/vmind/blob/main/LICENSE) +**Recommended: use the Skills directly in Codex or Claude Code.** Your agent provides model access and tool execution. Other compatible agents may also use these Skills, but do not have dedicated setup instructions here. -
- Edit src/App.tsx and save to test HMR
-
Click on the Vite and React logos to learn more
- - ); -} - -export default App; diff --git a/common/autoinstallers/create/templates/library/vite/src/assets/react.svg b/common/autoinstallers/create/templates/library/vite/src/assets/react.svg deleted file mode 100644 index 6c87de9b..00000000 --- a/common/autoinstallers/create/templates/library/vite/src/assets/react.svg +++ /dev/null @@ -1 +0,0 @@ - \ No newline at end of file diff --git a/common/autoinstallers/create/templates/library/vite/src/index.css b/common/autoinstallers/create/templates/library/vite/src/index.css deleted file mode 100644 index 2c3fac68..00000000 --- a/common/autoinstallers/create/templates/library/vite/src/index.css +++ /dev/null @@ -1,69 +0,0 @@ -:root { - font-family: Inter, system-ui, Avenir, Helvetica, Arial, sans-serif; - line-height: 1.5; - font-weight: 400; - - color-scheme: light dark; - color: rgba(255, 255, 255, 0.87); - background-color: #242424; - - font-synthesis: none; - text-rendering: optimizeLegibility; - -webkit-font-smoothing: antialiased; - -moz-osx-font-smoothing: grayscale; - -webkit-text-size-adjust: 100%; -} - -a { - font-weight: 500; - color: #646cff; - text-decoration: inherit; -} -a:hover { - color: #535bf2; -} - -body { - margin: 0; - display: flex; - place-items: center; - min-width: 320px; - min-height: 100vh; -} - -h1 { - font-size: 3.2em; - line-height: 1.1; -} - -button { - border-radius: 8px; - border: 1px solid transparent; - padding: 0.6em 1.2em; - font-size: 1em; - font-weight: 500; - font-family: inherit; - background-color: #1a1a1a; - cursor: pointer; - transition: border-color 0.25s; -} -button:hover { - border-color: #646cff; -} -button:focus, -button:focus-visible { - outline: 4px auto -webkit-focus-ring-color; -} - -@media (prefers-color-scheme: light) { - :root { - color: #213547; - background-color: #ffffff; - } - a:hover { - color: #747bff; - } - button { - background-color: #f9f9f9; - } -} diff --git a/common/autoinstallers/create/templates/library/vite/src/main.tsx b/common/autoinstallers/create/templates/library/vite/src/main.tsx deleted file mode 100644 index b3ad8a59..00000000 --- a/common/autoinstallers/create/templates/library/vite/src/main.tsx +++ /dev/null @@ -1,11 +0,0 @@ -import React from 'react'; -import ReactDOM from 'react-dom'; -import App from './App.tsx'; -import './index.css'; - -ReactDOM.render( -
-
-## Obtain DeepSeek API Key
-
-Register with DeepSeek and create your API key on the [DeepSeek API](https://platform.deepseek.com/api_keys) official website.
-
-
-
-## Configure Cursor
-
-Official website: https://www.cursor.com/
-
-Download and register, open your VChart project with Cursor, and configure Cursor.
-
-Using DeepSeek-V3 as an example, its API model name is `deepseek-chat`, and the API address is https://api.deepseek.com/v1. For more details, see the [API usage official website](https://api-docs.deepseek.com/zh-cn).
-
-Create a new model on the model page, set the corresponding API address and model name.
-
-
-
-Of course, you can also use any other AI model; here, we use DeepSeek.
-
-# Inject Official Tutorial into @Docs
-
-# Practical Verification
-
-## VTable Test
-
-VTable (https://www.visactor.io/vtable/; https://www.visactor.com/vtable/) is a powerful table component in the VisActor visualization library. It is designed to meet diverse data presentation needs, offering high flexibility and customizability. Whether it's simple data listing or complex data analysis display scenarios, VTable can provide excellent solutions.
-
-VTable has the following core features:
-
-- Supports various table types: basic tables, pivot tables, transposed tables, pivot charts, etc.
-
-- Powerful interaction features: sorting, filtering, row and column dragging, cell editing, etc.
-
-- Rich cell types: text, charts, progress bars, checkboxes, sparklines, etc.
-
-- High-performance rendering: supports smooth display of millions of data
-
-- Multi-platform adaptation: perfect support for mainstream frameworks like Vue, React
-
-### Invoke AI Interaction Panel in Cursor and Generate Code
-
-After setting up the project environment, use `cmd + i` to invoke the AI interaction panel. We directly let AI generate a basic table code snippet for inserting VTable.
-
-
-
-However, the initial generation result shows that AI does not recognize VTable's `ListTable`, possibly due to the inability to accurately recognize VTable's ListTable configuration.
-
-
-
-### Inject Official Tutorial into @Docs
-
-To solve the above problem, we inject the official tutorial into @Docs.
-
-
-
-Explicitly specify `VisActor VTable` in the `prompt` in @Docs. After this step, we are pleasantly surprised to find that AI can correctly write the implementation logic according to the `option` in VTable.
-
-
-
-### Code Application and Effect Display
-
-Copy the generated code into the corresponding file, run the project, and you can see the initial effect, with the table correctly generated.
-
-
-
-Then, we continue to let AI modify the table style. AI provided reasonable modification suggestions,
-
-
-
-After running the project again, we got a display effect that better meets the requirements:
-
-
-
-## VChart Test
-
-### Add a Bar Chart
-
-Invoke AI interaction with the `cmd+i` command, and let AI help us generate a simple bar chart code first.
-
-
-
-Directly apply this spec, and we check the result; a simple bar chart is rendered. It can be seen that DeepSeek has a certain understanding of VChart, and simple charts can be directly added. Let's try a more complex scenario.
-
-
-
-### Complex Scenario, Inject Docs
-
-We hope to add an average auxiliary line on the y-axis, check the result, but the result is incorrect. Upon closer inspection, it can be found that although the markLine is written as if it is correct, the spec does not conform to the specification, and the average line is calculated. We solve this problem by injecting docs.
-
-
-
-
-
-#### Set Docs
-
-Enter the Cursor settings page, select `Features`, add new docs, the docs address is https://visactor.com/vchart (https://visactor.io/vchart); you can also directly add through @Docs on the editing page.
-
-
-
-### Experimental Results
-
-By editing again with the newly added docs, the correct result can be obtained!
-
-
-
-
-
-## VStory Test
-
-VStory (GitHub: https://github.com/VisActor/VStory/; site: https://www.visactor.io/vstory/, https://www.visactor.com/vstory/) is a narrative-oriented visualization development framework that integrates the capabilities of all VisActor visualization components, making it more challenging to use. We conduct a simple test.
-
-Invoke AI interaction with the `cmd+i` command, and let AI help us generate a simple dashboard demo first.
-
-
-
-It can be found that it is completely incorrect because DeepSeek uses data from 2023, and VStory had not been released at that time, so it does not know how to use it. At this point, we need to let it read the documentation to learn.
-
-### Inject Docs
-
-We solve this problem by injecting docs.
-
-Enter the Cursor settings page, select `Features`, add new docs, the docs address is https://visactor.com/vstory/guide/tutorial_docs/VStory_Website_Guide; you can also directly add through @Docs on the editing page.
-
-
-
-### Experimental Results
-
-By editing again with the newly added docs, the correct result can be obtained!
-
-
-
-
-
-**Since VStory uses VChart, VTable, and VRender, to achieve better results, you should add the documentation of VChart, VTable, and VRender to the context simultaneously.**
-
-# Simple Summary
-
-1. Improve Development Efficiency
-
-- Quick Access: Directly access VisActor API documentation through Cursor
-
-- Intelligent Suggestions: Precise code completion based on VisActor documentation
-
-2. Enhance Development Experience
-
-- Seamless Integration: Directly obtain VTable usage guidance in the development environment
-
-- Real-time Feedback: Quickly verify code effects
-
-3. Reduce Learning Costs
-
-- Documentation Assistance: Access VisActor official examples at any time
-
-- Code Generation: Automatically generate code that complies with VisActor standards
-
-4. Improve Code Quality
-
-- Standard Check: Ensure code complies with VTable best practices
-
-- Performance Optimization: Automatically generate high-performance table configurations
-
-# Contact Us
-
-GitHub: [github.com/VisActor](https://link.juejin.cn/?target=https%3A%2F%2Fgithub.com%2FVisActor)
-
-Leave a message on the VisActor WeChat subscription account (you can join the WeChat group through the subscription account menu):
-
-
-
-VisActor official website: [www.visactor.io/](https://link.juejin.cn/?target=https%3A%2F%2Fwww.visactor.io%2Fvtable); [www.visactor.](https://link.juejin.cn/?target=https%3A%2F%2Fwww.visactor.io%2Fvtable)com
-
-Feishu Group:
-
-
-
-Discord: https://discord.com/invite/3wPyxVyH6m
-
-
----
-
-The doc is contributed by [玄魂](https://github.com/xuanhun)
\ No newline at end of file
diff --git a/docs/assets/guide/en/Intro_to_VMind.md b/docs/assets/guide/en/Intro_to_VMind.md
deleted file mode 100644
index 4c349da7..00000000
--- a/docs/assets/guide/en/Intro_to_VMind.md
+++ /dev/null
@@ -1,73 +0,0 @@
-
-# Introduction to VMind
-## Preface
-VisActor is a full-process visualization solution, which takes "visual storytelling" and "intelligence" as its core development direction. The components of VisActor that carry the ability of "visual storytelling" mainly include VChart, VTable, and VGrammar, all of which provide simple and easy-to-use interfaces. However, we hope that in some typical scenarios, users can generate charts and narrative works without writing code, thereby reducing the learning cost and complexity for users. The powerful generation ability of the large language model provides a natural language interaction interface for the VisActor visualization system, allowing us to directly call various functions of visualization components such as VChart and VTable through natural language, thereby completing visualization creation simply, quickly, and with high quality.
-
-
-
-
-
-VMind is an intelligent visualization component built on the basic capabilities of visualization component libraries such as VChart and VTable. We hope that in some typical scenarios, users can generate charts and narrative works with just one sentence, thereby reducing the learning cost and complexity for users.
-
-## VMind Architecture Design
-Intelligence is the future core direction of VisActor. Only on this basis can we truly reduce the access and development costs of developers and allow them to focus on things outside of visualization. VMind is an intelligent visualization component in the VisActor system. It provides intelligent interfaces in various ways such as rule algorithms, artificial intelligence, machine learning, and large language models to minimize the mental burden of developers using visualization components.
-
-
-
-
-
-In the VMind component, the bottom layer is the chart rendering, animation, annotation capabilities, VChart documentation provided by the VChart chart library, and the natural language understanding and code generation capabilities provided by the LLM. Among them, the data processing and interaction layer is responsible for processing user input data and natural language, including data aggregation modules for field parsing, data conversion, aggregation, etc., and the interaction communication module is responsible for handling communication between users and VMind, VMind and the LLM, including user intent recognition, Prompt management, model communication, result parsing.
-
-## VMind Core Capabilities
-
-
-
-
-
-The core capabilities of VMind include intelligent chart generation, intelligent chart editing, and intelligent color matching.
-- Intelligent chart generation refers to the process of directly generating charts from the display intent + data described in natural language.
-- Intelligent chart editing is the process of generating edited charts based on the original chart through the editing intent described in natural language.
-- Intelligent color matching is the process of generating stylized charts from natural language.
-
-Advanced capabilities are further applications of core capabilities, including complete visual narrative generation, data interpretation, and insight reports.
-
-## VMind Chart Creation Process
-
-
-
-
-
-Based on VMind's intelligent chart generation, conversational editing, and intelligent color matching modules, we can summarize the full process of using VMind for chart creation:
-First, the user provides a natural language description of the data display intent and chart data, or provides a natural language description of the chart editing intent and data news fragments, etc. These inputs are processed by the intent recognition module and routed to different modules. The intelligent chart generation module generates preliminary charts; the intelligent color matching module will generate corresponding theme color palettes according to user input, making the color matching of the chart more harmonious and beautiful; the conversational editing module will repeatedly apply a series of editing operations to fine-tune and optimize the chart until a chart that meets user needs is generated.
-
-The collaboration of these modules allows users to create visualization works without any threshold. This process not only greatly simplifies the complexity of chart creation but also greatly improves the efficiency and quality of chart creation.
-
-This process makes full use of the function call capability of the LLM. Chart generation, chart editing, and intelligent color matching are provided as functions for the model to call. LLM serves as the interface for interacting with users, selects the corresponding function according to user intent, and passes in parameters to execute the function of the corresponding module. In the future, VMind will also integrate insight extraction and data interpretation functions to interpret the generated charts, generate insight reports, or add annotations to the charts to display insight information in the data. At the same time, the model also has a powerful natural language dialogue capability, which can have multiple rounds of dialogue with users, and help users complete the entire process from chart creation to data analysis in a simple, natural, and friendly manner. Developers can easily integrate VMind into their systems as a data visualization assistant.
-
-## Practical Application of VMind
-### Data Video
-Data video is a type of video that presents and interprets data content through charts, animations, annotations, voiceovers, etc. This method can transform abstract data into intuitive graphics, present it in a more lively and interesting way, and make it easier for the audience to understand and accept.
-
-The steps to make a data video include:
-
-- Determine the data source and theme
-- Collect and organize data
-- Create, edit, and export charts
-- Video editing
-In the traditional data video production process, after collecting data, the creator needs to use professional visualization software to convert the data into charts; or use various chart libraries to draw charts through programming and export them as videos. In the era of short videos, everyone can become a video creator. However, non-professional video creators often do not have the skills to visualize data into charts and export them as videos, which greatly increases the threshold for making data videos.
-
-When using VMind to make data videos, users only need to describe the content they want to show in one sentence after collecting the data, and they can directly generate chart animations and support one-click export to videos and GIFs, which greatly reduces the difficulty of making data videos.
-
-
-
-
-
-### Intelligent Chart Assistant
-With the intelligent chart assistant built with VMind, you can use the function call capability of the large language model to judge the user's intent and call the VMind interface to realize chart generation and data interpretation, thereby reducing the difficulty of data visualization.
-
-
-__You can try the VMind Chart Assistant on [Cici](https://ciciai.com/bot/fUuxvgvv).__
-
-
-
-
diff --git a/docs/assets/guide/en/VMind_OpenAPI.md b/docs/assets/guide/en/VMind_OpenAPI.md
deleted file mode 100644
index fefafe3f..00000000
--- a/docs/assets/guide/en/VMind_OpenAPI.md
+++ /dev/null
@@ -1,90 +0,0 @@
-# VMind OpenAPI
-
-Built on the VMind chart generation tool, the VisActor team has launched VMind Agent, which improves the accuracy and efficiency of intelligent generation and adds multi‑turn editing capabilities. VMind Agent currently provides OpenAPI interfaces that you can call to achieve more advanced intelligent generation. The [VMind Agent Playground](/vmind/playground) offers a simple testing environment to experience VMind Agent’s capabilities.
-
-## VSeed
-
-VSeed is a domain‑specific language (DSL) for analytics. It encapsulates commonly used charting capabilities, greatly improving the efficiency of building analytic reporting platforms. In most cases, given a dataset and chart type, it can render directly; each chart type comes with extensive default conventions.
-
-VSeed is a subset of VChart and VTable. Its configuration is greatly simplified compared with VChart and VTable, making the options easier for large models to understand. It is a chart library specifically packaged for LLM‑driven generation.
-
-VSeed repository: [https://github.com/VisActor/VSeed](https://github.com/VisActor/VSeed)
-
-VSeed showcase: [https://visactor.github.io/VSeed/galley/chartType/line.html](https://visactor.github.io/VSeed/galley/chartType/line.html)
-
-VMind Agent prioritizes using VSeed for code generation for better results, while also supporting native code generation for VChart and VTable.
-
-## VMind Agent
-
-The overall technical approach of VMind Agent is as follows:
-
-When the user provides instructions and code blocks to VMind Agent, the agent uses the LLM to actively call tools and proceeds step by step through Generate/Edit → Check → Edit → Check … until the user’s requirements are met, outputting the corresponding code blocks and explanatory text.
-
-Compared with traditional chart generation tools, VMind Agent offers advantages in the following areas:
-- 🤖 ReAct architecture: Reasoning–action loop implemented with LangGraph
-- 🛠️ Diverse tools: Supports multiple chart generation and editing tools
-- 📊 Chart types: Supports many chart and table types
-- 🎨 Template system: Prebuilt chart templates
-- 🔧 Chart editing: Supports modifying existing charts
-- 💬 Natural language: Supports generating charts from pure natural‑language input
-
-### VMind Agent Architecture
-
-VMind Agent is composed of a Coordinator (scheduler) and a set of tools, and is implemented with the `@langchain/langgraph` library. The main modules are:
-
-
-
-#### Coordinator
-
-The scheduler’s main job is to determine, based on the current state, which tool to call next or whether to end the task.
-
-#### Generate Tool
-
-The generation tool is invoked when the user has not provided code or explicitly asks to regenerate code. If no chart type is specified, the tool recommends a suitable chart type based on the data. Each chart or table type provides a basic template; the LLM uses the user’s data and the template to generate the simplest code block for subsequent editing.
-
-#### Edit Tool
-
-The editing tool is the core of VMind Agent. Based on the current code and the user’s goal, the LLM breaks the task into multiple simple steps and completes them one by one. Before executing each edit step, the tool retrieves relevant materials from the knowledge base according to the step’s requirements to assist the LLM.
-
-#### Check Tool
-
-After each edit operation, the check tool verifies whether the change introduces serious errors or fails to achieve the step’s goal. If the check fails, the previous edit is rolled back.
-
-#### Short‑term Memory
-
-Each conversation records the current `sessionId`, allowing the same user to recall the state saved in previous turns and continue multi‑turn editing.
-
-### VMind Agent Knowledge Base
-
-To prevent hallucinations during editing, VMind Agent strictly limits the edit scope to the knowledge base fragments provided for each operation. The knowledge base consists of two parts:
-
-#### RAG
-
-Tutorial documents, demo examples, and Q&A content have strong independence and are well suited to a RAG approach. Relevant fragments are vectorized and stored in a vector database. For each edit operation, the system retrieves a set of related texts and code snippets for the model’s reference. When editing, if highly similar content is retrieved, the model can accurately perform the corresponding edits on the matched code snippets.
-
-#### Type Docs
-
-Configuration schemas for charts and tables are split by chart/table type and by module (e.g., axes, legends) into separate typed documentation fragments. Each fragment includes property names, types, and descriptions, and different fragments are indexed by different TopKeys. The LLM summarizes each fragment into Markdown and stores it in the database. For each TopKey, the LLM also produces a concise summary describing the scope covered by that TopKey.
-
-For each edit, the system first retrieves relevant TopKeys based on the instruction, then injects the associated Markdown fragments into the prompt. The LLM’s edits are strictly constrained to the range covered by the retrieved knowledge to avoid hallucination‑driven errors.
-
-## VMind Agent Applications
-
-### Playground
-
-VMind Agent Playground provides a simple testing environment to experience the agent’s capabilities. You can enter instructions and code blocks to see the generation and editing results.
-
-[VMind Agent Playground](/vmind/playground)
-
-### Douyin Mini Program
-
-Based on VMind Agent, we have developed a sample mini program. In Douyin mini programs, search for “数据智能洞察助手”. Users can input text for intelligent insights and experience the agent’s generation and editing results.
-
-
-
-## VMind OpenAPI Support
-
-Join our Feishu group to contact us and consult on VMind OpenAPI usage.
-
-
-
diff --git a/docs/assets/guide/en/VMind_Website_Guide.md b/docs/assets/guide/en/VMind_Website_Guide.md
deleted file mode 100644
index 6a788e72..00000000
--- a/docs/assets/guide/en/VMind_Website_Guide.md
+++ /dev/null
@@ -1,54 +0,0 @@
-
-# VMind Site Guide
-
-The purpose of this article is to help you use the VMind website more effectively and quickly find the information you need.
-
-## VMind OpenAPI
-
-VMind provides OpenAPI interfaces that allow you to achieve more advanced intelligent generation capabilities by calling these interfaces. You can learn more about OpenAPI in the [VMind OpenAPI](./VMind_OpenAPI) section.
-
-## Quick Start
-
-### Getting Started
-
-First, you need to read the [Getting Started](./Getting_Started) section of VMind. This section will teach you how to configure and use the environment required by VMind, as well as how to use VMind to build your first intelligent visualization application.
-
-### Learn More About VMind
-
-After you complete [Getting Started](./Getting_Started), you can learn about the functional positioning and related concepts of VMind in [Intro to VMind](./Intro_to_VMind). The [Create VMind Object](./Create_VMind_Object) chapter will teach you how to initialize a VMind object for subsequent operations. You can also learn about the data types supported by VMind in the [Data Input and Process](./Data_Input_and_Process) chapter. In the [Chart Generation](./Chart_Generation) chapter, you can learn more about how to use VMind for intelligent chart generation. The [Data Aggregation](./Data_Aggregation) chapter will teach you how to sort, filter, and aggregate data using VMind.
-
-## Documentation
-
-The VMind documentation provides detailed information on functions and APIs. Depending on your needs, you can view the following sections:
-
-- [Tutorials](./Basic_Tutorial/Create_VMind_Instance): Introduces the basic concepts of VMind and various usage methods.
-
-- [API](../api/VMind_Instance): Provides detailed descriptions of all available interfaces of VMind.
-
-
-
-## How to Use the Search Function
-
-This site provides a powerful search function that allows you to quickly find relevant usage information. Click on the search box at the top of the site, enter the keyword, and you will see the matching results in the drop-down list. Select the appropriate result to view the relevant content.
-
-## How to Ask Questions & Suggestions
-
-We are happy to help you! If you encounter any problems while using VMind, you can ask questions in the following ways:
-
-1. Submit an issue in the GitHub repository: Visit [VMind GitHub](https://github.com/VisActor/VMind/issues/new/choose), describe the problem you encountered in detail, and our team will respond and resolve it as soon as possible.
-
-2. Submit a discussion in the GitHub repository: Visit [VMind Discussion](https://github.com/VisActor/VMind/discussions), and we welcome you to share your ideas and suggestions here. Our team will respond and resolve it as soon as possible.
-
-## How to Report Errors
-
-If you find any problems in the document examples or think that some parts can be improved, please let us know. You can provide us with error correction information in the following ways:
-
-1. Submit a pull request in the GitHub repository: Correct the content and submit it, and a team member will review and merge it.
-
-2. Submit an issue: Point out the problems in the document, and a team member will verify and correct it as soon as possible.
-
-Thank you for your help! We will continue to improve the documentation and provide a better learning experience for all VMind users.
diff --git a/docs/assets/guide/menu.json b/docs/assets/guide/menu.json
deleted file mode 100644
index 4ee20fce..00000000
--- a/docs/assets/guide/menu.json
+++ /dev/null
@@ -1,128 +0,0 @@
-{
- "menu": "scripts",
- "children": [
- {
- "path": "VMind_Website_Guide",
- "title": {
- "zh": "VMind站点指引",
- "en": "VMind Website Guide"
- }
- },
- {
- "path": "VMind_OpenAPI",
- "title": {
- "zh": "VMind OpenAPI",
- "en": "VMind OpenAPI"
- }
- },
- {
- "path": "Intro_to_VMind",
- "title": {
- "zh": "VMind简介",
- "en": "Introduction to VMind"
- }
- },
- {
- "path": "Getting_Started",
- "title": {
- "zh": "快速上手",
- "en": "Getting Started"
- },
- "children": [
- {
- "path": "Getting_Started",
- "title": {
- "zh": "VMind快速上手",
- "en": "Getting Started with VMind"
- }
- },
- {
- "path": "Quickly_Get_Started_with_VisActor_Using_Trae_Context_Doc",
- "title": {
- "zh": "使用Trace Context快速上手VisActor",
- "en": "Quickly Get Started with VisActor Using Trace Context"
- }
- }
- ]
- },
- {
- "path": "Contribution_Guide",
- "title": {
- "zh": "贡献指南",
- "en": "Contribution Guide"
- }
- },
- {
- "path": "Basic_Tutorial",
- "title": {
- "zh": "基础教程",
- "en": "Basic Tutorial"
- },
- "children": [
- {
- "path": "How_to_Get_VMind",
- "title": {
- "zh": "如何获取VMind",
- "en": "How to Get VMind"
- }
- },
- {
- "path": "How_to_Import_VMind",
- "title": {
- "zh": "如何在项目中引用 VMind",
- "en": "How to Import VMind"
- }
- },
- {
- "path": "Create_VMind_Instance",
- "title": {
- "zh": "创建VMind实例",
- "en": "Create VMind Instance"
- }
- },
- {
- "path": "Data_Process",
- "title": {
- "zh": "数据格式与数据处理",
- "en": "Data Format and Data Process"
- }
- },
- {
- "path": "Data_Aggregation",
- "title": {
- "zh": "智能数据聚合",
- "en": "Intelligent Data Aggregation"
- }
- },
- {
- "path": "Chart_Generation",
- "title": {
- "zh": "图表智能生成",
- "en": "Intelligent Chart Generation"
- }
- },
- {
- "path": "Chart_Advisor",
- "title": {
- "zh": "基于规则的图表生成",
- "en": "Rule-based Chart Generation"
- }
- },
- {
- "path": "Data_Extraction",
- "title": {
- "zh": "数据提取——从文本一步生成图表",
- "en": "Text to Chart"
- }
- },
- {
- "path": "Chart_Insight",
- "title": {
- "zh": "智能洞察",
- "en": "Auto Insight"
- }
- }
- ]
- }
- ]
-}
\ No newline at end of file
diff --git a/docs/assets/guide/zh/Basic_Tutorial/Chart_Advisor.md b/docs/assets/guide/zh/Basic_Tutorial/Chart_Advisor.md
deleted file mode 100644
index 0da2e0ce..00000000
--- a/docs/assets/guide/zh/Basic_Tutorial/Chart_Advisor.md
+++ /dev/null
@@ -1,84 +0,0 @@
-# 基于规则的图表生成
-VMind可以结合内建的@visactor/chart-advisor，自动根据当前数据集生成图表。chart-advisor完整依赖于可视化规则，不需要调用大型语言模型。当你使用generateChart，而大模型无法成功生成图表时，VMind会使用chart-advisor作为备选方案来创建一个图表。
-
-你同样可以主动选择使用chart-advisor来生成图表，只需在初始化VMind对象时，将模型类型设置为chart-advisor：
-```ts
-import VMind, { Model } from '@visactor/vmind'
-
-const vmind = new VMind({
-model: Model.CHART_ADVISOR,
-})
-const userPrompt = '';
-const advisorResult = await vmind.generateChart(userPrompt, fieldInfo, dataset);
-```
-由于chart-advisor不依赖大型模型，userPrompt可以设为空。
-你可以在[数据格式与数据处理](./Data_Process)章节了解fieldInfo和dataset的更多相关信息。
-
-chart-advisor会根据一系列规则生成几种可用的图表，此时generateChart将返回包含图表类型、得分以及spec的列表。
-
-## 示例
-下面是一个使用商品销售额数据，通过chart-advisor生成图表的示例：
-
-```ts
-const dataset = [
-{
-"Product name": "Coke",
-"Sales": 2350
-},
-{
-"Product name": "Sprite",
-"total_sales": 1056
-},
-{
-"Product name": "Fanta",
-"total_sales": 4778
-},
-{
-"Product name": "Mirinda",
-"total_sales": 3336
-}
-// ...more data
-]
-
-const fieldInfo = [
-{
-"fieldName": "Product name",
-"type": "string",
-"role": "dimension"
-},
-{
-"fieldName": "Sales",
-"type": "int",
-"role": "measure"
-}
-]
-
-const advisorResult = await vmind.generateChart(userPrompt, fieldInfo, dataset);
-
-```
-advisorResult结果如下：
-```json
-[
-{
-"chartSource": "chartAdvisor", //图表来源
-"spec": ..., //生成的图表spec
-"chartType": "BAR CHART", //图表类型
-"score": 1, //当前图表得分
-},
-{
-"chartSource": "chartAdvisor",
-"spec": ..., //生成的图表spec
-"chartType": "RADAR CHART", //图表类型
-"score": 0.16666666666666666 //当前图表得分
-}
-]
-```
-
-其中，score表示当前图表类型与输入数据的匹配程度，分数越高表示该类型的图表越适用于展示当前数据。
-
-
-生成的图表如下：
-
-
-
-根据输入数据的特性以及fieldInfo中的字段信息，chart-advisor会基于一整套可视化规则，为每种图表类型打分，并依分数将各图表类型排序。这些规则确保了可视化结果的准确度和美观性，并尽可能地减少图表中的视觉混乱和视觉错误。
diff --git a/docs/assets/guide/zh/Basic_Tutorial/Chart_Generation.md b/docs/assets/guide/zh/Basic_Tutorial/Chart_Generation.md
deleted file mode 100644
index 309fa323..00000000
--- a/docs/assets/guide/zh/Basic_Tutorial/Chart_Generation.md
+++ /dev/null
@@ -1,545 +0,0 @@
-# 图表生成
-本教程将向你详细介绍VMind中的图表智能生成功能，并提供一些示例。
-
-生成图表的方式有很多，比如你可以在PowerBI、Tableau等专业的BI可视化工具中，利用数据集的字段来制作可视化图表；或者你也可以直接使用VChart、ECharts、MatPlotlib等图表库，通过编写代码来绘制图表。此外，VChart、Echarts等图表库还提供了简单易用的图表编辑器，用户可以上传数据并进行图表制作。
-
-然而，传统的图表生成方式也存在一些问题：
-
-- 对于编程新手来说，他们可能不会编写代码，也就无法使用图表库来生成图表。
-- 图表类型繁多，用户展示数据的目的各不相同，这就导致了适合展示的图表类型和对应的字段映射也会有所不同。对于非专业用户来说，他们可能无法准确地选择出最能表达自己观点的图表类型。
-
-- 用户对于图表样式、动画的需求各不相同，往往需要一定的设计能力才能满足。设计出令人眼前一亮的图表并不容易，它需要设计技能和审美意识，而这些是许多用户所缺乏的。
-
-这些问题增加了数据可视化的创作门槛，阻碍了非专业用户创作符合自己意图的可视化作品。
-
-VMind的解决方案是，利用大语言模型的自然语言理解能力和公域知识，根据用户的数据和展示意图，推荐合适的图表类型，并将字段映射到合适的视觉通道上，实现图表的智能生成，从而降低用户进行数据可视化的门槛。
-
-
-## generateChart
-VMind的generateChart函数是一个强大的工具，它可以帮助你智能生成图表。这个函数需要以下几个参数：
-- userPrompt (string): 用户的可视化意图（你想用图表展示什么信息）
-- fieldInfo (Array): 数据集中字段的信息，包括字段名称，类型等
-- dataset (Array): 图表中使用的原始数据集，可以是undefined。若dataset为undefined，则会生成一个spec模板，后续可以调用fillSpecTemplateWithData将数据填充到spec模板。
-- options: 可选，选项参数，包括以下内容：
-
- - chartTypeList (ChartType[]，可选): 支持的图表类型列表。若不为undefined，则会从该列表指定的图表类型中选择生成。
- - enableDataQuery (boolean, 可选): 决定是否在图表生成过程中开启数据聚合
- - colorPalette (Array
-
-# 什么上下文Doc 功能
-
-指定上下文可以让 AI 助手提供更精准的回答，但是大模型对一个组件的理解的实时程度是低于组件自己的官方文档的。如果我们在使用一个不熟悉的组件的时候，如果能将该组件的相关文档引入到上下文中，可以大大的提升智能化编程水平。
-
-
-
-Trae 使用 “#” 来指定上下文。
-
-
-
-输入“#Doc：”，可以索引文档集。 我们添加VisActor 的文档，进行测试。 目前Trae 还没有内置任何文档集，所以需要手动添加。
-
-首先添加文档集：
-
-
-
-
-
-这里我们通过url的方式添加各组件的入口url：
-
-* VChart：https://www.visactor.io/vchart/
-
-* VTable：https://www.visactor.io/vtable
-
-* VMind：https://www.visactor.io/vmind
-
-* ....
-
-
-
-整个过程略微缓慢，但是索引效果，明显好于 Cursor。
-
-
-
-
-
-下面我们实地测试一下引用文档作为上下文是否会给编程带来实际的益处。
-
-# 效果测试
-
-## 创建测试项目
-
-我们创建一个最简单web 项目
-
-```
-npm create vite@latest my-ts-web -- --template typescript
-cd my-ts-web
-
-```
-
-
-# 实践验证
-
-## 新增一个线图
-
-我们要求Trae 生成一段代码，在当前文件中添加一个VChart 线图。
-
-
-
-生成的结果还是有错误的。
-
-
-
-我们下面引入文档，看看是不是会有所改变。
-
-
-
-重新生成的代码修复了之前的api 错误调用。
-
-
-
-生成了一个正确的线图。
-
-
-
-## 加大难度
-
-首先我们来修改数据，显示三条线。
-
-
-
-问题的分析和答案都比较靠谱：
-
-
-
-
-
-主要修改点：
-
-1. 重构了数据结构，使用type字段来区分不同系列
-
-1. 使用seriesField来指定系列的分类字段
-
-1. 每个数据点包含year（x轴）、value（y轴）和type（系列）三个字段
-
-
-
-修改入场动画效果：
-
-
-
-
-
-主要修改：
-
-1. 添加了animation配置
-
-1. appear设置入场动画为fadeIn（淡入淡出）
-
-1. duration设置动画持续时间为1000毫秒
-
-结果如下：
-
-
-
-
-
-## 更多问答
-
-因为有了官方文档，我们可以让Trae 回答更多类型的问题，不只是代码本身。
-
-比如基本的图表概念：
-
-
-
-或者 具体用法：
-
-
-
-导航到更相关文档：
-
-
-
-
-
-# 简单总结
-
-Trae 结合文档有如下好处：
-
-1. 提升开发效率
-
-- 快速访问：通过Marscode直接查阅VisActor API文档
-
-- 智能提示：基于VisActor文档的精准代码补全
-
-2. 增强开发体验
-
-- 无缝集成：在开发环境中直接获取VTable使用指导
-
-- 实时反馈：快速验证代码效果
-
-3. 降低学习成本
-
-- 文档辅助：随时查阅VisActor官方示例
-
-- 代码生成：自动生成符合VisActor规范的代码
-
-4. 提高代码质量
-
-- 规范检查：确保代码符合VisActor最佳实践
-
-- 性能优化：自动生成高性能表格配置
-
-# 联系我们
-
-github ：[github.com/VisActor](https://link.juejin.cn/?target=https%3A%2F%2Fgithub.com%2FVisActor)
-
-VisActor 微信订阅号留言（可以通过订阅号菜单加入微信群）：
-
-
-
-VisActor 官网：[www.visactor.io/](https://link.juejin.cn/?target=https%3A%2F%2Fwww.visactor.io%2Fvtable) ； [www.visactor.](https://link.juejin.cn/?target=https%3A%2F%2Fwww.visactor.io%2Fvtable)com
-
-飞书群：
-
-
-
-discord：https://discord.com/invite/3wPyxVyH6m
-
-
----
-
-The doc is contributed by [玄魂](https://github.com/xuanhun)
diff --git a/docs/assets/guide/zh/Intro_to_VMind.md b/docs/assets/guide/zh/Intro_to_VMind.md
deleted file mode 100644
index 1ca6bbd4..00000000
--- a/docs/assets/guide/zh/Intro_to_VMind.md
+++ /dev/null
@@ -1,86 +0,0 @@
-# VMind简介
-## 前言
-VisActor是一款全流程可视化解决方案，它以“可视化叙事”和“智能化”为核心发展方向。VisActor承载“可视化叙事” 能力的组件主要有VChart、VTable 和 VGrammar，它们都提供了简单易用的接口。然而，我们希望在一些典型场景中，用户能够无需编写代码就能生成图表和叙事作品，以此降低用户的学习成本和使用复杂度。大语言模型的强大生成能力为VisActor可视化体系提供了一个自然语言的交互接口，使我们能够通过自然语言直接调用VChart、VTable等可视化组件的各项功能，从而简单、快速、高质量地完成可视化创作。
-
-
-
-
-
-VMind是在VChart、VTable等可视化组件库的基础能力之上，构建的一个智能可视化组件。我们希望在一些典型场景中，用户能够只需一句话就能生成图表和叙事作品，以此降低用户的学习成本和使用复杂度。
-
-## VMind架构设计
-智能化是VisActor的未来核心方向，只有在此基础上，我们才能真正降低开发者的接入和研发成本，让他们能够将注意力集中在可视化之外。VMind是VisActor体系中的智能可视化组件，它基于规则算法、人工智能、机器学习、大语言模型等多种方式提供智能化接口，以最大限度降低开发者使用可视化组件的心智负担。
-
-
-
-
-
-在VMind组件中，最底层是由VChart图表库提供的图表渲染、动画、标注能力、VChart文档，以及大语言模型提供的自然语言理解、代码生成能力。其中，数据处理与交互层负责处理用户输入的数据和自然语言，包括数据聚合模块进行字段解析、数据转换、聚合等操作，交互通信模块负责处理用户与VMind、VMind与大语言模型之间的通信，包括用户意图识别、Prompt管理，模型通信，结果解析。
-
-
-## VMind核心能力
-
-
-
-
-
-VMind的核心能力包括图表智能生成、图表智能编辑和智能配色。
-- 图表智能生成，是指从自然语言描述的展示意图+数据直接生成图表的过程
-- 图表智能编辑是在原有图表的基础上，通过自然语言描述编辑意图，生成编辑后的图表
-- 智能配色是从自然语言生成风格化图表的过程。
-
-进阶能力是对核心能力的进一步应用，包括完整的可视化叙事生成、数据解读与洞察报告。
-
-
-## VMind图表创作流程
-
-
-
-
-
-基于VMind的图表智能生成、对话式编辑和智能配色模块，我们可以概括出使用VMind进行图表创作的全流程：
-首先，用户提供自然语言描述的数据展示意图和图表数据，或者提供自然语言描述的图表编辑意图和数据新闻片段等。这些输入经过意图识别模块的处理，路由到不同模块。图表智能生成模块生成初步的图表；智能配色模块会根据用户的输入生成相应的主题色板，使得图表的颜色搭配更加和谐、美观；对话式编辑模块会反复应用一系列编辑操作，对图表进行微调和优化，直到生成一张满足用户需求的图表。
-
-这些模块之间的协同作用，使得用户可以零门槛创作可视化作品。这种流程不仅大大简化了图表创作的复杂度，也极大地提升了图表创作的效率和质量。
-
-这一流程充分利用了大模型的function call能力。图表生成、图表编辑与智能配色作为function供模型调用，LLM作为与用户交互的接口，根据用户意图选择对应的function并传入参数，执行对应模块的功能。在未来，VMind还将集成洞察提取、数据解读功能，对生成的图表进行解读，生成洞察报告，或者在图表中添加标注展示数据中的洞察信息。与此同时，模型还具有强大的自然语言对话能力，能够与用户进行多轮对话，简单、自然、友好地帮助用户完成从图表创作到数据分析的全过程。开发人员可以轻松地将VMind集成到他们的系统中，作为一个数据可视化助手。
-
-
-
-## VMind使用场景
-### 数据视频
-数据视频是一种将数据内容通过图表、动画、标注、旁白等形式进行展示和解读的视频类型。这种方式能够将原本抽象的数据转化为直观的图形，以更加生动有趣的方式呈现，使观众更容易理解和接受。
-
-制作数据视频的步骤包括：
-
-- 确定数据来源和主题
-- 收集和整理数据
-- 制作、编辑和导出图表
-- 视频编辑
-在传统的数据视频制作流程中，创作者需要在收集数据后，使用专业的可视化软件将数据转化为图表；或者使用各种图表库，通过编程绘制图表并导出为视频。在短视频时代，每个人都有可能成为视频的创作者。但是，非专业的视频创作者往往没有将数据可视化为图表并导出为视频的技能，这就大大提高了制作数据视频的门槛。
-
-而使用 VMind 制作数据视频，用户只需要在收集好数据之后，用一句话描述想要展示的内容，就能直接生成图表动画，并支持一键导出为视频和 GIF，这极大地降低了制作数据视频的难度。
-
-
-
-
-
-
-### 图表智能助手
-借助VMind打造的图表智能助手，你可以利用大语言模型的function call能力，判断用户的意图并调用VMind接口，实现图表的生成和数据的解读，从而降低数据可视化的难度。
-
-
-__你可以在[Cici](https://ciciai.com/bot/fUuxvgvv)体验VMind图表助手。__
-
-
-
-
-
-
-
-
-
-
-
-
-
diff --git a/docs/assets/guide/zh/VMind_OpenAPI.md b/docs/assets/guide/zh/VMind_OpenAPI.md
deleted file mode 100644
index 18846485..00000000
--- a/docs/assets/guide/zh/VMind_OpenAPI.md
+++ /dev/null
@@ -1,90 +0,0 @@
-# VMind OpenAPI
-
-在VMind图表生成工具的基础上，VisActor团队全新推出了VMind Agent，提升了智能生成的准确性和效率，并新增了多轮编辑能力。目前VMind Agent提供了OpenAPI接口，你可以通过调用这些接口来实现更高级的智能生成能力。[VMind Agent Playground](/vmind/playground)提供了一个简单的测试环境，方便体验VMind Agent的能力。
-
-## VSeed
-
-VSeed是一个面向分析领域的DSL, 封装了该领域常用的图表能力, 极大地提升分析式报表平台的搭建效率。一般情况下, 给定数据集和图表类型, 即可直接出图，每类图表类型存在大量的默认约定逻辑。
-
-VSeed是VChart和VTable的子集，配置相比VChart和VTable大幅简化，配置项更适合大模型理解，是专为LLM生成封装的图表库。
-
-VSeed 项目仓库：[https://github.com/VisActor/VSeed](https://github.com/VisActor/VSeed)
-
-VSeed 能力展示：[https://visactor.github.io/VSeed/galley/chartType/line.html](https://visactor.github.io/VSeed/galley/chartType/line.html)
-
-VMind Agent优先推荐使用VSeed进行代码生成，生成效果更加优秀，也同时支持VChart和VTable的原生代码生成。
-
-## VMind Agent
-
-VMind Agent的整体技术方案如下：
-
-用户输入的指令和代码块，输入VMind Agent后，Agent会基于LLM主动调用工具，开始逐步进行 生成/编辑 -> 检查 -> 编辑 -> 检查 ...... 任务流，最终完成用户的要求，输出对应的代码块和文本解释说明。
-
-相比传统的图表生成工具，VMind Agent在一下方面有优势：
-- 🤖 ReAct 架构: 基于 LangGraph 实现的推理-行动循环
-- 🛠️ 多样化工具: 支持多种图表生成和编辑工具
-- 📊 图表类型: 支持多种图表类型和表格类型
-- 🎨 模板系统: 预置多种图表模板
-- 🔧 图表编辑: 支持对现有图表进行修改
-- 💬 自然语言: 支持纯自然语言输入生成图表
-
-### VMind Agent 架构
-
-VMind Agent 主要由Coordinator调度器和若干工具组成，基于`@langchain/langgraph库`完成Agent的实现。主要模块如下：
-
-
-
-#### Coordinator
-
-调度器主要工作是：基于当前状态，判断下一步应该调用哪一个工具，或结束任务。
-
-#### Generate Tool
-
-生成工具在用户没有输入代码，或指令要求重新生成代码时调用。如果没有指定图表类型，工具会基于数据推荐适合的图表类型。每一种图表/表格类型都会提供基础的模版，LLM会基于用户的数据和模版生成最简单的代码块，供进一步编辑使用。
-
-#### Edit Tool
-
-编辑工具是VMind Agent的核心部分，LLM会基于当前的代码和用户的目标，将任务拆分为多个简单的步骤并逐一完成。在执行每次编辑任务前，工具会基于本次步骤的任务要求，在知识库中召回对应的资料，辅助LLM进行编辑。
-
-#### Check Tool
-
-检查工具在每一步编辑操作后执行，用来检查当次编辑操作是否会引起恶性错误，或是否没有达到当次步骤的目标。如果检查不通过，会回退上一次编辑操作。
-
-#### 短期记忆
-
-每一次对话会记录当前的sessionId，方便同一个用户进行多轮对话时，找回之前轮次的对话所记录的状态，方便用户进行多轮编辑操作。
-
-### VMind Agent 知识库
-
-为了避免LLM在编辑代码时出现幻觉，VMind Agent将编辑范围严格限制在每次编辑操作提供的知识库片段中。知识库主要分为两部分：
-
-#### RAG
-
-教程文档、示例Demo和Q&A问答内容，内容独立性强，适合使用RAG方案。相应片段进行向量化后，存入向量数据库。每次进行编辑操作时，会召回对应的若干文本及代码片段，供大模型参考。
-大模型进行编辑时，如果召回有非常接近的内容，会已经对应的代码片段准确进行相应的编辑。
-
-#### Type Docs
-
-图表和表格的配置Schema，按照不同的图表/表格类型，以及不同的模块（例如坐标轴、图例等）进行拆分，组织成一个个类型说明片段。每个片段中有该部分对应的属性名称、属性类型和属性说明，不同的片段使用不同的TopKey进行索引。使用LLM将每个每个片段进行归纳总结，生成对应的说明Markdown片段，存储在数据库中。每一个TopKey也依据对应的Markdown片段，使用LLM总结一段简单的说明，介绍这个TopKey下的功能范围。
-
-每次编辑时，首先依据编辑指令召回对应的若干TopKey，之后将TopKey对应的Markdown文本片段，作为知识库加入prompt中，大模型的编辑操作会被严格限制在本次召回的知识库的范围，避免幻觉影响编辑结果。
-
-## VMind Agent 应用
-
-### Playground
-
-VMind Agent Playground提供了一个简单的测试环境，方便体验VMind Agent的能力。你可以在Playground中输入指令和代码块，查看Agent的生成和编辑结果。
-
-[VMind Agent Playground](/vmind/playground)
-
-### 抖音小程序
-
-基于VMind Agent的，我们开发了一个示例小程序，抖音小程序搜索 数据智能洞察助手 对话小程序，用户可以在小程序中输入需要进行智能洞察的文本，体验Agent的生成和编辑结果。
-
-
-
-## VMind OpenAPI 使用咨询
-
-加入飞书群欢迎联系我们，咨询VMind OpenAPI的使用方案。
-
-
diff --git a/docs/assets/guide/zh/VMind_Website_Guide.md b/docs/assets/guide/zh/VMind_Website_Guide.md
deleted file mode 100644
index c712c20f..00000000
--- a/docs/assets/guide/zh/VMind_Website_Guide.md
+++ /dev/null
@@ -1,55 +0,0 @@
-
-# VMind站点指南
-
-本篇文章的目的是为了帮助你更有效地使用 VMind 网站，让你能更快地找到你需要的信息。
-
-## VMind OpenAPI
-
-VMind 提供了 OpenAPI 接口，你可以通过调用这些接口来实现更高级的智能生成能力。你可以在[VMind OpenAPI](./VMind_OpenAPI)章节了解更多关于 OpenAPI 的信息。
-
-## 快速开始
-
-### 快速上手
-
-首先，你需要阅读 VMind 的[快速上手](./Getting_Started)章节。这个章节会教你如何配置和使用 VMind 所需的环境，以及如何使用 VMind 构建你的第一个智能可视化应用。
-
-### 深入了解 VMind
-
-在你完成[快速上手](./Getting_Started)后，你可以在[VMind简介](./Intro_to_VMind)了解 VMind 的功能定位和相关概念。
-[创建VMind对象](./Create_VMind_Object)章节将教你如何初始化一个 VMind 对象以进行后续操作。你还可以通过阅读[数据处理与输入](./Data_Input_and_Process)章节来了解 VMind 支持的数据类型。在[图表生成](./Chart_Generation)章节，你可以深入了解如何使用 VMind 进行图表智能生成。[数据聚合](./Data_Aggregation)章节将教你如何使用 VMind 对数据进行排序、筛选、聚合等操作。
-
-## 文档
-
-VMind 的文档提供了详细的功能和 API 信息。根据你的需求，你可以查看以下几个部分：
-
-- [教程](./Basic_Tutorial/Create_VMind_Instance)：介绍了 VMind 的基本概念和各种使用方法。
-
-- [API](../api/VMind_Instance)：提供了 VMind 的所有可用接口的详细说明。
-
-
-
-## 如何使用搜索功能
-
-本站点提供了一个强大的搜索功能，让你可以快速找到相关的使用信息。点击站点顶部的搜索框，输入关键词后，你将在下拉列表中看到匹配的结果。选择合适的结果即可查看相关内容。
-
-## 如何提问&建议
-
-我们很高兴为你提供帮助! 如果你在使用 VMind 过程中遇到问题，可以通过以下方式提问：
-
-1. 在 GitHub 仓库中提交一个 issue：访问 [VMind GitHub](https://github.com/VisActor/VMind/issues/new/choose)，详细描述你遇到的问题，我们的团队将尽快回应并解决。
-
-2. 在 GitHub 仓库提交一个 discussion：访问 [VMind Discussion](https://github.com/VisActor/VMind/discussions)，非常欢迎你在这里提出你的想法和建议，我们的团队将尽快回应并解决。
-
-## 如何纠错
-
-如果你在文档示例中发现问题，或者认为某些部分可以改进，务必告诉我们。你可以通过以下方式为我们提纠错信息：
-
-1. 在 GitHub 仓库中提交一个 pull request：纠正的内容并提交，会有团队成员审核并合并。
-
-2. 提交一个 issue：指出文档中存在的问题，团队成员会尽快核实并修正。
-
-感谢你的帮助！我们将不断完善文档，为所有 VMind 用户提供更好的学习体验。
diff --git a/docs/assets/openManus/Amazon_Fashion_Sales_Analysis_Report.md b/docs/assets/openManus/Amazon_Fashion_Sales_Analysis_Report.md
deleted file mode 100644
index fbfca095..00000000
--- a/docs/assets/openManus/Amazon_Fashion_Sales_Analysis_Report.md
+++ /dev/null
@@ -1,125 +0,0 @@
-# Amazon Fashion Sales Analysis and Growth Strategy
-## January 2025 Performance Review and 10% Growth Plan
-
-## Executive Summary
-
-This comprehensive analysis examines the January 2025 sales data from your Amazon Fashion store to identify patterns, trends, and opportunities for growth. The analysis reveals several key insights, including day-of-week performance variations, the impact of discount strategies, and customer purchasing behaviors. Based on these findings, we've developed a multi-faceted strategy projected to increase sales by 14%, well above the requested 10% target. The strategy focuses on optimizing day-specific promotions, enhancing customer purchase behavior, refining pricing and promotion tactics, and improving inventory management. This report provides both detailed analysis and actionable recommendations with implementation timelines to achieve sustainable growth.
-
-## Data Analysis Methodology
-
-The analysis began with a thorough exploration of the January 2025 sales data, which included 31 days of daily metrics covering orders, customers, units sold, and various financial indicators. The data was preprocessed to create derived metrics such as average order value, units per order, profit margins, and day-of-week features. Comprehensive statistical analysis was performed to identify patterns, correlations, and anomalies in the data. Multiple visualizations were created to illustrate key trends and relationships, providing a foundation for data-driven strategy development.
-
-## Sales Performance Overview
-
-January 2025 showed overall healthy performance with a total of 1,213 orders, generating approximately $166,000 in gross sales and $57,000 in net profit. The average daily performance included 39 orders, 101 units sold, and $5,349 in gross sales, with an average order value of $136.11 and a profit margin of 34.11%. The data revealed significant day-to-day variations, with a notable peak on January 14th that substantially outperformed all other days.
-
-### Weekly Performance Trends
-
-The sales data demonstrated a clear pattern of growth and decline throughout the month. Week 3 (January 15-21) showed peak performance with 316 orders, $45,546 in gross sales, and $15,827 in net profit. This represented a 39.8% increase in orders and a 49.9% increase in gross sales compared to Week 1. However, Weeks 4 and 5 showed a progressive decline, with Week 5 recording only 148 orders and $20,751 in gross sales. This mid-month peak followed by end-of-month decline suggests a cyclical pattern that requires strategic intervention to maintain momentum throughout the entire month.
-
-Average order value also fluctuated weekly, peaking in Week 3 at $144.97, while Week 4 had the lowest AOV at $125.07. This 16% variation in AOV between weeks indicates significant potential for optimization through targeted promotional strategies. Profit margins showed similar fluctuations, ranging from 31.64% in Week 4 to 36.44% in Week 5, suggesting opportunities for margin improvement through strategic product mix and pricing adjustments.
-
-### Day-of-Week Analysis
-
-The analysis revealed substantial performance variations by day of week. Tuesdays consistently outperformed other days with an average of 52.25 orders and $7,152 in gross sales, while Thursdays showed the weakest performance with only 28.6 orders and $3,763 in gross sales. This 82.7% difference in order volume between the best and worst-performing days represents a significant opportunity for targeted interventions to boost underperforming days.
-
-Weekend days (Saturday and Sunday) averaged higher orders (43.88) compared to weekdays (37.48), but showed lower average order values ($129.74 vs. $138.33). This suggests different shopping behaviors between weekday and weekend customers, with weekend shoppers making more frequent but smaller purchases. Mondays showed the highest average order value at $146.42, indicating potential for premium product focus on this day.
-
-### Outlier Analysis: January 14th
-
-Tuesday, January 14th stood out as an exceptional outlier, with 79 orders (82.31% above the average Tuesday), $11,752 in gross sales (109.14% above average), and $4,010 in net profit (113.58% above average). This single day accounted for approximately 7% of the month's total sales. The exceptional performance extended across all metrics, including a higher average order value ($148.76 vs. $128.29) and more units per order (2.84 vs. 2.42) compared to other Tuesdays.
-
-The dramatic outperformance of January 14th suggests specific factors that drove success on this day, potentially including promotional activities, external events, or seasonal factors. Identifying and replicating these success factors represents a significant opportunity for boosting overall sales performance.
-
-## Customer Behavior Insights
-
-The analysis revealed several important patterns in customer purchasing behavior that provide opportunities for optimization. The near-perfect 100% conversion rate across all days suggests most visitors are returning customers or highly qualified leads, indicating a loyal customer base but also potential for expanding reach to new customers.
-
-Customers purchased an average of 2.57 units per order, with higher units per order on Tuesdays (2.84) and lower on weekends (2.35). This variation suggests different shopping motivations throughout the week and opportunities for increasing basket size through strategic cross-selling and bundling. The data also showed that orders and unique customers were nearly identical, indicating most customers place only one order per day, highlighting potential for encouraging multiple purchases.
-
-## Financial Performance Analysis
-
-Profit margins showed significant variation, ranging from 26.97% to 42.03%, with an average of 34.11%. Weekdays demonstrated higher profit margins (35.47%) compared to weekends (33.38%), suggesting different product mix preferences between weekday and weekend shoppers. The correlation analysis revealed a moderate positive relationship (0.41) between profit margin and average order value, indicating that higher-value orders tend to be more profitable.
-
-The cost structure analysis showed Amazon fees remained relatively consistent at 19-21% of gross sales, while product costs varied more significantly, ranging from 38.39% to 50.67% of gross sales. This variation in product costs suggests opportunities for optimizing the product mix to favor higher-margin items during strategic promotional periods.
-
-## Discount and Shipping Impact
-
-The analysis of discount strategies revealed that medium discounts ($1-2 per order) correlated with the highest orders (44.17), gross sales ($6,174.53), and net profit ($2,129.66). Higher discounts showed diminishing returns, with lower profit margins (32.09% for high discounts vs. 34.60% for low discounts) and no corresponding increase in order volume. Discounts showed negative correlations with average order value (-0.10) and profit margin (-0.32), suggesting the need for strategic application to maximize overall profitability.
-
-Shipping charge analysis showed that medium shipping charges ($5-6 per order) correlated with the highest profit margins (35.47%). Shipping charges showed minimal correlation with orders (-0.01) and sales (0.01), suggesting limited price sensitivity to shipping costs within the current range. The negative correlation (-0.39) between shipping and discount suggests these are used as alternative promotional tactics, providing an opportunity for more strategic coordination between these levers.
-
-## Data-Driven Growth Strategies
-
-Based on the comprehensive analysis of January 2025 sales data, we've developed a multi-faceted strategy to achieve a minimum 10% increase in sales for the upcoming month. The strategy focuses on four key areas: day-of-week optimization, customer behavior optimization, pricing and promotion refinement, and inventory management improvements.
-
-### Day-of-Week Optimization Strategies
-
-The significant performance variations by day of week present clear opportunities for targeted interventions. For Tuesdays, which already show strong performance, we recommend enhancing their success by analyzing what made January 14th exceptional and implementing similar tactics on all Tuesdays. This includes implementing 3-hour flash sales with medium-level discounts ($1-2 per order) that showed optimal performance in our analysis, and scheduling targeted email campaigns to be delivered Monday evening to drive Tuesday traffic. These Tuesday enhancements are projected to increase Tuesday sales by 15%, contributing 3% to overall monthly sales.
-
-For underperforming Thursdays, we recommend a rescue plan featuring Thursday-exclusive promotions, free shipping offers, and bundle deals to increase units per order from the current 2.57 average. These Thursday interventions are expected to increase Thursday sales by 20%, contributing 2% to overall monthly sales.
-
-Weekend strategies should focus on improving average order value, which currently lags behind weekday performance. This includes promoting higher-value items on weekends and implementing tiered shipping discounts that increase with order value to encourage larger weekend purchases. These weekend AOV improvements are projected to increase weekend AOV by 8%, contributing 1.5% to overall monthly sales.
-
-### Customer Behavior Optimization
-
-To increase units per order from the current 2.57 average, we recommend implementing AI-powered product recommendations based on cart contents, offering tiered discounts for purchasing multiple items, and featuring complementary products on product pages and during checkout. These strategies are expected to increase average units per order to 2.8 (+9%), contributing 2% to overall monthly sales.
-
-To encourage repeat purchases within the same month, we recommend sending follow-up emails with time-limited discount codes for next purchases and implementing a points-based loyalty program. These repeat purchase incentives are projected to generate 5% additional orders from repeat customers, contributing 1% to overall monthly sales.
-
-### Pricing and Promotion Optimization
-
-Based on the analysis of discount impact, we recommend refining the discount strategy to focus on the medium discount range ($1-2 per order) that showed highest orders, gross sales, and net profit. Discounts should be applied strategically to products with higher profit margins to maintain overall profitability. This optimized discount strategy is expected to improve conversion while maintaining profit margins, contributing 1.5% to overall monthly sales.
-
-For shipping strategy, we recommend implementing free shipping for orders above $150 to increase average order value (currently $136.11) and focusing on medium shipping charges ($5-6) that showed highest profit margins. These shipping optimizations are projected to increase AOV by 5%, contributing 1% to overall monthly sales.
-
-### Inventory and Product Mix Strategies
-
-To optimize the product mix, we recommend identifying and featuring products with above-average profit margins and expanding selection in categories showing highest units per order. This product mix optimization is expected to improve overall profit margin while increasing sales, contributing 1% to overall monthly sales.
-
-To address the observed decline in sales during Weeks 4-5, we recommend creating special promotions for the last week of the month and implementing a strategic discounting of slow-moving inventory during the final week. These end-of-month boost strategies are projected to reduce the Week 4-5 decline by 50%, contributing 1% to overall monthly sales.
-
-## Implementation Plan
-
-The implementation of these strategies should follow a phased approach over four weeks:
-
-Week 1 should focus on building the foundation by setting up technical infrastructure for all promotions, preparing email templates and marketing materials, and configuring discount rules and shipping tiers.
-
-Week 2 should launch the core strategies, including the Tuesday Power Day strategy, Units Per Order initiatives, and Discount Strategy Refinement.
-
-Week 3 should expand the implementation to include the Thursday Rescue Plan, Weekend AOV Improvement tactics, and Repeat Purchase Incentives.
-
-Week 4 should focus on optimization by deploying the End-of-Month Sales Boost, fine-tuning all strategies based on early results, and preparing for continuation in the following month.
-
-## Expected Results
-
-The combined impact of these strategies is projected to increase sales by 14%, exceeding the 10% target:
-- Day-of-Week Optimization: +6.5%
-- Customer Behavior Optimization: +3.0%
-- Pricing and Promotion Optimization: +2.5%
-- Inventory and Product Mix Strategies: +2.0%
-
-This comprehensive approach provides a buffer for any strategies that may underperform expectations while maintaining or improving profit margins. The strategies are designed to be sustainable and scalable, providing a foundation for continued growth beyond the immediate target period.
-
-## Conclusion
-
-The January 2025 sales data reveals a healthy business with significant opportunities for optimization and growth. By strategically addressing day-of-week performance variations, enhancing customer purchase behavior, refining pricing and promotion tactics, and improving inventory management, we project a 14% increase in sales for the upcoming month. The recommended strategies are data-driven, actionable, and designed to not only increase sales volume but also maintain or improve profit margins. Implementation should follow the phased approach outlined in this report, with ongoing monitoring and adjustment to maximize results.
-
-## Appendix: Key Visualizations
-
-The analysis includes several key visualizations that illustrate the patterns and insights discussed in this report:
-
-1. Daily Gross Sales Trend - January 2025
-2. Weekly Gross Sales - January 2025
-3. Average Daily Orders by Day of Week
-4. Weekend vs Weekday Performance Comparison
-5. January 14th vs Average Tuesday Comparison
-6. Impact of Discount Levels on Sales Metrics
-7. Profit Margin by Discount Range
-8. Units per Order vs Average Order Value
-9. Impact of Shipping Levels on Sales Metrics
-10. Daily Profit Margin Trend - January 2025
-11. Daily Average Order Value Trend - January 2025
-12. Correlation Matrix of Sales Metrics
-
-These visualizations provide visual confirmation of the patterns identified in the analysis and support the strategic recommendations outlined in this report.
diff --git a/docs/assets/openManus/Readme.md b/docs/assets/openManus/Readme.md
deleted file mode 100644
index c5a8bb65..00000000
--- a/docs/assets/openManus/Readme.md
+++ /dev/null
@@ -1,3 +0,0 @@
-# Case of Sales Analysis and Strategies for Amazon Store
-
-The current folder contains all execution artifacts of [the Sales Analysis and Strategies for Amazon Store case in manus](https://manus.im/share/c3onakN6Iajcm1Vt1xAVG7?replay=1)
\ No newline at end of file
diff --git a/docs/assets/openManus/amazon_sales_jan2025.csv b/docs/assets/openManus/amazon_sales_jan2025.csv
deleted file mode 100644
index 177fcadd..00000000
--- a/docs/assets/openManus/amazon_sales_jan2025.csv
+++ /dev/null
@@ -1,32 +0,0 @@
-Date,Orders,Unique Customers,Units Sold,Gross Sales,Shipping Revenue,Coupon Discounts,Amazon Fees,Product Costs,Net Profit
-2025-01-01,37,37,83,4813.44,275.91,706.65,734.33,2241.35,1407.02
-2025-01-02,43,43,111,6182.74,489.61,833.44,848.23,2936.95,2053.73
-2025-01-03,32,31,114,6914.35,427.26,888.52,1074.84,3491.21,1887.05
-2025-01-04,41,41,91,6138.23,471.18,643.65,747.82,3200.84,2017.1
-2025-01-05,46,46,102,7394.39,509.73,966.82,903.9,3815.65,2217.75
-2025-01-06,32,32,116,9157.3,676.22,1369.04,1550.84,4377.13,2536.5
-2025-01-07,35,35,86,4861.82,316.4,520.24,688.49,2652.06,1317.44
-2025-01-08,38,38,99,5039.76,339.21,634.59,643.19,2639.77,1461.42
-2025-01-09,41,40,114,3188.5,206.54,458.72,482.53,1611.58,842.2
-2025-01-10,35,34,101,7134.65,398.52,977.72,1185.78,3647.03,1722.64
-2025-01-11,38,37,92,7297.66,523.05,984.09,970.22,3590.15,2276.24
-2025-01-12,33,33,96,4047.84,236.49,546.96,637.02,1921.81,1178.54
-2025-01-13,41,41,96,6275.02,318.33,740.29,785.14,3047.13,2020.79
-2025-01-14,41,41,107,7174.17,497.63,822.73,883.15,3772.07,2193.85
-2025-01-15,79,78,93,11752.01,650.04,1650.78,1784.91,5305.32,3661.04
-2025-01-16,42,42,95,6619.66,517.75,930.1,1009.09,3055.68,2142.54
-2025-01-17,30,30,101,5610.81,441.11,804.33,887.89,2550.68,1809.03
-2025-01-18,34,33,94,3523.28,272.86,513.21,576.29,1599.82,1106.82
-2025-01-19,35,35,93,6922.58,423.0,869.25,1236.03,3707.36,1532.94
-2025-01-20,40,39,93,7336.22,370.21,917.58,1107.61,3817.52,1863.72
-2025-01-21,51,50,109,6450.31,502.15,902.49,899.03,3208.5,1942.45
-2025-01-22,41,40,108,3916.81,246.15,518.97,656.89,1800.89,1186.22
-2025-01-23,29,29,105,4221.79,333.52,570.36,575.22,2107.36,1302.38
-2025-01-24,24,24,102,3652.92,288.25,510.64,534.56,1816.77,1079.19
-2025-01-25,37,37,112,6503.53,491.6,939.76,811.04,3039.23,2205.1
-2025-01-26,46,46,91,7251.14,426.61,847.66,881.17,3577.61,2371.32
-2025-01-27,35,35,104,5985.29,368.41,710.93,1063.94,2931.9,1646.94
-2025-01-28,31,31,92,6759.36,510.56,707.7,1150.17,3457.99,1954.07
-2025-01-29,44,44,90,4047.1,240.83,521.73,654.65,2078.22,1033.33
-2025-01-30,43,42,98,5567.35,306.68,566.74,804.69,2530.53,1972.07
-2025-01-31,44,44,88,3293.75,219.71,406.05,429.5,1605.58,1072.33
diff --git a/docs/assets/openManus/aov_trend.png b/docs/assets/openManus/aov_trend.png
deleted file mode 100644
index 31cafa78..00000000
Binary files a/docs/assets/openManus/aov_trend.png and /dev/null differ
diff --git a/docs/assets/openManus/correlation_heatmap.png b/docs/assets/openManus/correlation_heatmap.png
deleted file mode 100644
index c152b15d..00000000
Binary files a/docs/assets/openManus/correlation_heatmap.png and /dev/null differ
diff --git a/docs/assets/openManus/daily_sales_trend.png b/docs/assets/openManus/daily_sales_trend.png
deleted file mode 100644
index 935de153..00000000
Binary files a/docs/assets/openManus/daily_sales_trend.png and /dev/null differ
diff --git a/docs/assets/openManus/home/ubuntu/analysis/data_exploration.txt b/docs/assets/openManus/home/ubuntu/analysis/data_exploration.txt
deleted file mode 100644
index 1a1b9ad9..00000000
--- a/docs/assets/openManus/home/ubuntu/analysis/data_exploration.txt
+++ /dev/null
@@ -1,33 +0,0 @@
-Amazon Fashion Sales Data Exploration - January 2025
-
-Dataset Overview:
-- 31 rows (daily sales for January 2025)
-- 10 columns including date, orders, customers, and financial metrics
-- No missing values in the dataset
-
-Columns:
-- Date: Date of sales (datetime64)
-- Orders: Number of orders per day (int64)
-- Unique Customers: Number of unique customers per day (int64)
-- Units Sold: Total units sold per day (int64)
-- Gross Sales: Total sales amount before deductions (float64)
-- Shipping Revenue: Revenue from shipping (float64)
-- Coupon Discounts: Discount amount from coupons (float64)
-- Amazon Fees: Fees paid to Amazon (float64)
-- Product Costs: Cost of products sold (float64)
-- Net Profit: Final profit after all deductions (float64)
-
-Basic Statistics:
-- Average daily orders: ~39
-- Average daily unique customers: ~39
-- Average daily units sold: ~101
-- Average daily gross sales: ~$5,348.56
-- Average daily net profit: ~$1,835.56
-- Highest daily orders: 79
-- Highest daily gross sales: $11,752.01
-- Highest daily net profit: $4,010.05
-
-Initial Observations:
-- Orders and unique customers are nearly identical, suggesting most customers place only one order
-- There appears to be significant daily variation in sales metrics
-- Further analysis needed to identify patterns by day of week and other factors
diff --git a/docs/assets/openManus/home/ubuntu/analysis/preprocessing_results.txt b/docs/assets/openManus/home/ubuntu/analysis/preprocessing_results.txt
deleted file mode 100644
index aee77fe1..00000000
--- a/docs/assets/openManus/home/ubuntu/analysis/preprocessing_results.txt
+++ /dev/null
@@ -1,43 +0,0 @@
-## Preprocessing Results - Amazon Fashion Sales January 2025
-
-### Derived Metrics Summary
-- **Average Order Value**: Mean = $136.11, Min = $108.01, Max = $172.30
-- **Units per Order**: Mean = 2.57, Min = 2.11, Max = 3.15
-- **Profit Margin**: Mean = 34.11%, Min = 26.97%, Max = 42.03%
-- **Conversion Rate**: Mean = 100.13%, Min = 100.00%, Max = 102.56%
-- **Shipping per Order**: Mean = $5.99, Min = $3.88, Max = $7.92
-- **Discount per Order**: Mean = $1.36, Min = $0.43, Max = $2.99
-- **Amazon Fee Rate**: Mean = 19.96%, Min = 19.02%, Max = 20.85%
-- **Product Cost Rate**: Mean = 44.91%, Min = 38.39%, Max = 50.67%
-
-### Day of Week Distribution
-- Monday: 4 days
-- Tuesday: 4 days
-- Wednesday: 5 days
-- Thursday: 5 days
-- Friday: 5 days
-- Saturday: 4 days
-- Sunday: 4 days
-
-### Weekend vs Weekday Performance
-- **Weekday Orders**: 37.48 (average)
-- **Weekend Orders**: 43.88 (average)
-- **Weekday Gross Sales**: $5,217.34 (average)
-- **Weekend Gross Sales**: $5,725.82 (average)
-- **Weekday Net Profit**: $1,788.21 (average)
-- **Weekend Net Profit**: $1,971.69 (average)
-- **Weekday Average Order Value**: $138.33
-- **Weekend Average Order Value**: $129.74
-
-### Identified Outliers (Z-score > 2.5)
-- **January 14, 2025 (Tuesday)**:
- - Orders: 79 (significantly higher than average)
- - Gross Sales: $11,752.01 (significantly higher than average)
- - Net Profit: $4,010.05 (significantly higher than average)
-
-### Initial Observations
-1. Weekend sales are higher than weekday sales on average
-2. January 14th shows exceptional performance across multiple metrics
-3. Conversion rate is consistently near 100%, suggesting most visitors make purchases
-4. Profit margins vary significantly (26.97% to 42.03%)
-5. Average order value is higher on weekdays than weekends
diff --git a/docs/assets/openManus/home/ubuntu/analysis/sales_analysis_results.txt b/docs/assets/openManus/home/ubuntu/analysis/sales_analysis_results.txt
deleted file mode 100644
index 2fec0e66..00000000
--- a/docs/assets/openManus/home/ubuntu/analysis/sales_analysis_results.txt
+++ /dev/null
@@ -1,85 +0,0 @@
-# Amazon Fashion Sales Analysis - January 2025
-
-## Weekly Sales Summary
- Week Orders Gross Sales Net Profit Units Sold Average_Order_Value Profit_Margin
- 1 226 30386.58 9871.64 595 135.442626 32.708879
- 2 275 37780.80 13559.80 697 136.042737 35.675951
- 3 316 45545.99 15827.16 851 144.970516 34.343915
- 4 248 31341.21 10050.28 607 125.066667 31.643778
- 5 148 20750.81 7593.51 387 139.948597 36.441550
-
-## Day of Week Analysis
-Day_of_Week Orders Gross Sales Net Profit Units Sold Average_Order_Value Profit_Margin Conversion_Rate
- Monday 34.75 5232.0825 1872.7900 96.75 146.423938 35.822693 100.000000
- Tuesday 52.25 7152.3300 2410.6700 135.00 133.406436 33.274309 100.357143
- Wednesday 30.20 4313.9020 1537.0080 81.40 143.503032 35.605238 100.000000
- Thursday 28.60 3763.0660 1183.6480 73.20 131.742903 31.753670 100.000000
- Friday 44.00 6015.2760 2078.3560 113.60 137.213694 34.383283 100.000000
- Saturday 41.75 5397.4900 1805.6050 102.75 129.425496 33.383204 100.641026
- Sunday 46.00 6054.1400 2137.7675 114.50 130.048484 34.691468 100.000000
-
-## Correlation Analysis
- Orders Units Sold Gross Sales Net Profit Average_Order_Value Units_per_Order Profit_Margin Shipping_per_Order Discount_per_Order
-Orders 1.00 0.98 0.97 0.93 0.09 0.13 0.04 -0.01 0.21
-Units Sold 0.98 1.00 0.99 0.95 0.26 0.33 0.05 0.02 0.17
-Gross Sales 0.97 0.99 1.00 0.98 0.32 0.32 0.14 0.01 0.17
-Net Profit 0.93 0.95 0.98 1.00 0.38 0.31 0.33 -0.02 0.12
-Average_Order_Value 0.09 0.26 0.32 0.38 1.00 0.89 0.41 0.07 -0.10
-Units_per_Order 0.13 0.33 0.32 0.31 0.89 1.00 0.08 0.12 -0.12
-Profit_Margin 0.04 0.05 0.14 0.33 0.41 0.08 1.00 -0.04 -0.32
-Shipping_per_Order -0.01 0.02 0.01 -0.02 0.07 0.12 -0.04 1.00 -0.39
-Discount_per_Order 0.21 0.17 0.17 0.12 -0.10 -0.12 -0.32 -0.39 1.00
-
-## Outlier Analysis (January 14th)
- 13
-Date 2025-01-14 00:00:00
-Day_of_Week Tuesday
-Orders 79
-Units Sold 224
-Gross Sales 11752.01
-Net Profit 4010.05
-Average_Order_Value 148.75962
-Units_per_Order 2.835443
-Profit_Margin 34.122248
-Discount_per_Order 1.992278
-
-## Comparing January 14th (Tuesday) with average Tuesday
-Orders: Jan 14 = 79.00, Avg Tuesday = 43.33, Difference = 35.67 (82.31%)
-Units Sold: Jan 14 = 224.00, Avg Tuesday = 105.33, Difference = 118.67 (112.66%)
-Gross Sales: Jan 14 = 11752.01, Avg Tuesday = 5619.10, Difference = 6132.91 (109.14%)
-Net Profit: Jan 14 = 4010.05, Avg Tuesday = 1877.54, Difference = 2132.51 (113.58%)
-Average_Order_Value: Jan 14 = 148.76, Avg Tuesday = 128.29, Difference = 20.47 (15.96%)
-Units_per_Order: Jan 14 = 2.84, Avg Tuesday = 2.42, Difference = 0.41 (17.04%)
-Profit_Margin: Jan 14 = 34.12, Avg Tuesday = 32.99, Difference = 1.13 (3.43%)
-Discount_per_Order: Jan 14 = 1.99, Avg Tuesday = 1.64, Difference = 0.35 (21.50%)
-
-## Discount Impact Analysis
-Correlation between Discount_per_Order and other metrics:
-Discount_per_Order 1.000000
-Orders 0.205425
-Units Sold 0.170070
-Average_Order_Value -0.097984
-Units_per_Order -0.120498
-Gross Sales 0.173755
-Net Profit 0.117784
-
-Impact of discount levels:
-Discount_Range Orders Units Sold Gross Sales Net Profit Average_Order_Value Units_per_Order Profit_Margin
- Low (0-1) 31.111111 79.222222 4126.292222 1426.725556 132.357850 2.541760 34.598545
- Medium (1-2) 44.166667 116.444444 6174.531667 2129.664444 140.308968 2.637570 34.314201
- High (2-3) 34.500000 82.000000 4381.797500 1431.975000 125.683370 2.351389 32.088781
-
-## Shipping Revenue Impact Analysis
-Correlation between Shipping_per_Order and other metrics:
-Shipping_per_Order 1.000000
-Orders -0.008446
-Units Sold 0.019099
-Average_Order_Value 0.066474
-Units_per_Order 0.119193
-Gross Sales 0.005687
-Net Profit -0.020329
-
-Impact of shipping levels:
-Shipping_Range Orders Units Sold Gross Sales Net Profit Average_Order_Value Units_per_Order Profit_Margin
- Low (3-5) 36.200000 91.000000 4941.178000 1704.150000 135.848036 2.519047 33.325092
- Medium (5-6) 40.142857 102.714286 5400.252857 1911.112857 132.681980 2.509347 35.470619
\ No newline at end of file
diff --git a/docs/assets/openManus/home/ubuntu/todo.md b/docs/assets/openManus/home/ubuntu/todo.md
deleted file mode 100644
index 89e072b9..00000000
--- a/docs/assets/openManus/home/ubuntu/todo.md
+++ /dev/null
@@ -1,13 +0,0 @@
-# Amazon Fashion Sales Analysis Todo List
-
-## Data Preparation and Analysis
-- [x] Load and explore the Amazon Fashion Sales data
-- [x] Clean and preprocess data (create derived metrics)
-- [x] Perform comprehensive sales analysis
-- [x] Create data visualizations
-- [x] Identify key insights and trends
-
-## Strategy Development and Reporting
-- [x] Develop data-driven strategies to boost sales by 10%
-- [x] Prepare comprehensive analysis report
-- [x] Deliver final report with recommendations
diff --git a/docs/assets/openManus/key_insights.md b/docs/assets/openManus/key_insights.md
deleted file mode 100644
index 25354b86..00000000
--- a/docs/assets/openManus/key_insights.md
+++ /dev/null
@@ -1,60 +0,0 @@
-# Key Insights and Trends - Amazon Fashion Sales January 2025
-
-## Sales Patterns and Trends
-
-### Weekly Trends
-1. **Peak Performance in Week 3**: Week 3 (January 15-21) showed the highest performance with 316 orders, $45,545.99 in gross sales, and $15,827.16 in net profit.
-2. **Consistent Growth and Decline**: Sales showed growth from Week 1 to Week 3, followed by a decline in Weeks 4 and 5, suggesting a mid-month peak pattern.
-3. **Average Order Value Fluctuation**: AOV peaked in Week 3 at $144.97, while Week 4 had the lowest AOV at $125.07.
-
-### Day of Week Patterns
-1. **Tuesday Performance**: Tuesdays show the highest average orders (52.25) and gross sales ($7,152.33).
-2. **Weekend vs. Weekday**: Weekend days (Saturday and Sunday) average higher orders (43.88) compared to weekdays (37.48).
-3. **AOV Variation by Day**: Mondays have the highest average order value ($146.42), while Saturdays have the lowest ($129.43).
-4. **Thursday Underperformance**: Thursdays consistently show the lowest performance across orders, sales, and profit.
-
-### Outlier Analysis
-1. **January 14th Anomaly**: Tuesday, January 14th showed exceptional performance:
- - 79 orders (82.31% above average Tuesday)
- - $11,752.01 gross sales (109.14% above average)
- - $4,010.05 net profit (113.58% above average)
- - This single day accounted for approximately 7% of the month's total sales.
-
-## Customer Behavior Insights
-
-### Order Patterns
-1. **Conversion Rate Consistency**: Nearly perfect 100% conversion rate across all days, suggesting most visitors are returning customers or highly qualified leads.
-2. **Units per Order**: Average of 2.57 units per order, with highest on Tuesdays (2.84) and lowest on weekends (2.35).
-3. **One-time Purchasers**: Orders and unique customers are nearly identical, indicating most customers place only one order per day.
-
-## Financial Performance Insights
-
-### Profit Metrics
-1. **Profit Margin Variation**: Profit margins range from 26.97% to 42.03%, with an average of 34.11%.
-2. **Weekday vs. Weekend Profitability**: Weekdays show higher profit margins (35.47%) compared to weekends (33.38%).
-3. **Correlation with AOV**: Profit margin has a moderate positive correlation (0.41) with average order value.
-
-### Cost Structure
-1. **Amazon Fee Consistency**: Amazon fees remain relatively consistent at 19-21% of gross sales.
-2. **Product Cost Variation**: Product costs range from 38.39% to 50.67% of gross sales, suggesting variable product mix.
-
-## Discount and Shipping Impact
-
-### Discount Analysis
-1. **Medium Discount Sweet Spot**: Medium discounts (1-2$ per order) correlate with highest orders (44.17), gross sales ($6,174.53), and net profit ($2,129.66).
-2. **Discount-Profit Relationship**: Higher discounts correlate with lower profit margins (32.09% for high discounts vs. 34.60% for low discounts).
-3. **Negative Correlation**: Discounts show negative correlation with AOV (-0.10) and profit margin (-0.32).
-
-### Shipping Analysis
-1. **Medium Shipping Sweet Spot**: Medium shipping charges ($5-6 per order) correlate with highest profit margins (35.47%).
-2. **Minimal Impact**: Shipping charges show minimal correlation with orders (-0.01) and sales (0.01).
-3. **Discount-Shipping Relationship**: Negative correlation (-0.39) between shipping and discount, suggesting they may be used as alternative promotional tactics.
-
-## Opportunities for Growth
-
-1. **Tuesday Optimization**: Leverage the success factors from January 14th to replicate performance on other Tuesdays.
-2. **Weekend AOV Improvement**: Focus on increasing weekend AOV, which currently lags behind weekday performance.
-3. **Thursday Enhancement**: Develop specific strategies to boost underperforming Thursday sales.
-4. **Discount Strategy Refinement**: Optimize discount strategy around the 1-2$ per order range that shows best overall performance.
-5. **Units per Order Increase**: Potential to increase units per order, particularly on weekends.
-6. **Week 4-5 Performance**: Address the sales decline in the last two weeks of the month.
diff --git a/docs/assets/openManus/orders_by_day.png b/docs/assets/openManus/orders_by_day.png
deleted file mode 100644
index c5e6f03e..00000000
Binary files a/docs/assets/openManus/orders_by_day.png and /dev/null differ
diff --git a/docs/assets/openManus/outlier_comparison.png b/docs/assets/openManus/outlier_comparison.png
deleted file mode 100644
index d2dd6857..00000000
Binary files a/docs/assets/openManus/outlier_comparison.png and /dev/null differ
diff --git a/docs/assets/openManus/profit_margin_by_discount.png b/docs/assets/openManus/profit_margin_by_discount.png
deleted file mode 100644
index c94fa15a..00000000
Binary files a/docs/assets/openManus/profit_margin_by_discount.png and /dev/null differ
diff --git a/docs/assets/openManus/sales_by_day.png b/docs/assets/openManus/sales_by_day.png
deleted file mode 100644
index f7459790..00000000
Binary files a/docs/assets/openManus/sales_by_day.png and /dev/null differ
diff --git a/docs/assets/openManus/sales_growth_strategies.md b/docs/assets/openManus/sales_growth_strategies.md
deleted file mode 100644
index 7778e877..00000000
--- a/docs/assets/openManus/sales_growth_strategies.md
+++ /dev/null
@@ -1,90 +0,0 @@
-# Data-Driven Strategies to Boost Sales by 10%
-
-Based on the comprehensive analysis of January 2025 sales data, the following strategies are recommended to achieve a 10% increase in sales for the upcoming month:
-
-## 1. Day-of-Week Optimization Strategies
-
-### Tuesday Power Day Enhancement
-- **Replicate January 14th Success**: Analyze what made January 14th exceptional (82% more orders than average Tuesday) and implement similar tactics on all Tuesdays.
-- **Tuesday Flash Sales**: Implement 3-hour flash sales on Tuesdays with medium-level discounts ($1-2 per order) that showed optimal performance.
-- **Tuesday Email Campaign**: Schedule targeted email campaigns to be delivered Monday evening to drive Tuesday traffic.
-- **Expected Impact**: +15% increase in Tuesday sales, contributing +3% to overall monthly sales.
-
-### Thursday Rescue Plan
-- **Thursday-Exclusive Promotions**: Create Thursday-only promotions to boost the lowest-performing day of the week.
-- **Free Shipping Thursday**: Offer free shipping on Thursdays to overcome the lowest-performing day.
-- **Bundle Deals**: Introduce product bundles on Thursdays to increase units per order from current 2.57 average.
-- **Expected Impact**: +20% increase in Thursday sales, contributing +2% to overall monthly sales.
-
-### Weekend AOV Improvement
-- **Premium Weekend Focus**: Promote higher-value items on weekends to increase the weekend AOV (currently $129.74 vs. weekday $138.33).
-- **Tiered Shipping Discounts**: Implement shipping discounts that increase with order value to encourage larger weekend purchases.
-- **Expected Impact**: +8% increase in weekend AOV, contributing +1.5% to overall monthly sales.
-
-## 2. Customer Behavior Optimization
-
-### Increase Units Per Order
-- **Strategic Cross-Selling**: Implement AI-powered product recommendations based on cart contents.
-- **Bundle Pricing**: Offer tiered discounts for purchasing multiple items (e.g., 5% off for 2 items, 10% off for 3+ items).
-- **Complementary Product Highlights**: Feature complementary products on product pages and during checkout.
-- **Expected Impact**: Increase average units per order from 2.57 to 2.8 (+9%), contributing +2% to overall monthly sales.
-
-### Repeat Purchase Incentives
-- **Post-Purchase Discount Codes**: Send follow-up emails with time-limited discount codes for next purchase.
-- **Loyalty Program Launch**: Implement a points-based loyalty program to encourage repeat purchases within the same month.
-- **Expected Impact**: Generate 5% additional orders from repeat customers, contributing +1% to overall monthly sales.
-
-## 3. Pricing and Promotion Optimization
-
-### Discount Strategy Refinement
-- **Optimize Discount Levels**: Focus on the medium discount range ($1-2 per order) that showed highest orders (44.17), gross sales ($6,174.53), and net profit ($2,129.66).
-- **Targeted Discount Application**: Apply discounts strategically to products with higher profit margins to maintain overall profitability.
-- **Expected Impact**: Optimize conversion while maintaining profit margins, contributing +1.5% to overall monthly sales.
-
-### Shipping Strategy Enhancement
-- **Free Shipping Threshold**: Implement free shipping for orders above $150 to increase AOV (current average: $136.11).
-- **Shipping Tier Optimization**: Focus on medium shipping charges ($5-6) that showed highest profit margins (35.47%).
-- **Expected Impact**: Increase AOV by 5%, contributing +1% to overall monthly sales.
-
-## 4. Inventory and Product Mix Strategies
-
-### Product Mix Optimization
-- **Promote High-Margin Products**: Identify and feature products with above-average profit margins.
-- **Category Expansion**: Expand selection in categories showing highest units per order.
-- **Expected Impact**: Improve overall profit margin while increasing sales, contributing +1% to overall monthly sales.
-
-### End-of-Month Sales Boost
-- **Month-End Promotion**: Create special promotions for the last week of the month to address the observed decline in Weeks 4-5.
-- **Inventory Clearance Strategy**: Strategic discounting of slow-moving inventory during the final week.
-- **Expected Impact**: Reduce Week 4-5 decline by 50%, contributing +1% to overall monthly sales.
-
-## Implementation Timeline and Expected Results
-
-### Week 1: Foundation
-- Set up technical infrastructure for all promotions
-- Prepare email templates and marketing materials
-- Configure discount rules and shipping tiers
-
-### Week 2: Launch
-- Implement Tuesday Power Day strategy
-- Launch Units Per Order initiatives
-- Begin Discount Strategy Refinement
-
-### Week 3: Expansion
-- Roll out Thursday Rescue Plan
-- Implement Weekend AOV Improvement tactics
-- Launch Repeat Purchase Incentives
-
-### Week 4: Optimization
-- Deploy End-of-Month Sales Boost
-- Fine-tune all strategies based on early results
-- Prepare for next month continuation
-
-### Expected Overall Impact
-- Day-of-Week Optimization: +6.5%
-- Customer Behavior Optimization: +3.0%
-- Pricing and Promotion Optimization: +2.5%
-- Inventory and Product Mix Strategies: +2.0%
-- **Total Expected Sales Increase: +14.0%**
-
-This comprehensive approach exceeds the 10% sales increase target while maintaining or improving profit margins, providing a buffer for any strategies that may underperform expectations.
diff --git a/docs/assets/openManus/weekend_comparison.png b/docs/assets/openManus/weekend_comparison.png
deleted file mode 100644
index b3d84d26..00000000
Binary files a/docs/assets/openManus/weekend_comparison.png and /dev/null differ
diff --git a/docs/assets/openManus/weekly_sales.png b/docs/assets/openManus/weekly_sales.png
deleted file mode 100644
index 3209e44c..00000000
Binary files a/docs/assets/openManus/weekly_sales.png and /dev/null differ
diff --git a/docs/assets/scenario/en/Chart_Assistant.md b/docs/assets/scenario/en/Chart_Assistant.md
deleted file mode 100644
index 1c829697..00000000
--- a/docs/assets/scenario/en/Chart_Assistant.md
+++ /dev/null
@@ -1,15 +0,0 @@
-# Chart Intelligence Assistant
-VMind can serve as a visual agent for large language models, responsible for tasks such as chart generation and chart editing, thereby creating a "Chart Intelligence Assistant". This assistant can serve as an independent robot, or it can be integrated into other products in the form of a Copilot.
-
-
-
-
-
-
-As a chart intelligence assistant, VMind can generate corresponding charts based on the natural language description entered by the user. It can also realize secondary filtering, aggregation, and sorting of data through the intelligent data aggregation module.
-
-Whether you need to display a time series line chart or a bar chart of different categories, VMind can quickly respond and help you select the chart type and generate the chart. With VMind, users can create both aesthetically pleasing and information-rich charts without in-depth chart design knowledge, greatly improving work efficiency.
-
-## Experience address:
-- [Coze](https://www.coze.com/s/Zs8MNnSod/)
-- [Discord](https://discord.gg/sajqrVSP): Join the Discord channel and @VMind Chart Master bot experience in chat.
diff --git a/docs/assets/scenario/en/Data_Development.md b/docs/assets/scenario/en/Data_Development.md
deleted file mode 100644
index 4c0820e6..00000000
--- a/docs/assets/scenario/en/Data_Development.md
+++ /dev/null
@@ -1,57 +0,0 @@
-# Data Development
-In python Notebook scenarios such as data development, it is suitable to call VMind Open API to generate charts after SQL queries, and integrate intelligent visualization capabilities in Notebook with py-vchart rendering charts.
-
-
-
-
-
-## Sample Code
-```python
-'''
-In the Notebook scene, VMind Open API can be called to generate charts after SQL queries, and combined with py-vchart for chart rendering
-VMind supports chart generation based on large language models and rules
-The chart generation process will not pass detailed data to the large language model, so you can eat it with confidence
-'''
-
-import requests
-import pandas as pd
-from pyvchart import render_chart
-
-url = "" # VMind Open API url
-
-# Dataset queried:
-print('SQL query result: ')
-display(sales_data_df)
-
-sales_data_json =sales_data_df.to_json(orient="records", force_ascii=False)
-
-userPrompt = "Help me display sales of various products in different regions" # User's presentation intent
-
-print()
-print('Display intent input: ')
-print(userPrompt)
-print()
-
-body = {
-"data": sales_data_json,
-"userPrompt": userPrompt,
-"model": "doubao" # Use the doubao model
-}
-
-# Call VMind Open API to generate chart
-response = requests.post(url, headers=headers, json=body)
-
-chart = response.json().get('chart')[0]
-spec = chart.get('spec')
-chartType = chart.get('chartType')
-
-# print(spec)
-print('Generated chart type: '+chartType)
-
-# Call py-vchart to render chart
-render_chart(spec)
-```
-
-## Related Reading
-- [VMind Open API](./Open_API)
-- [py-vchart](https://visactor.io/vchart/guide/tutorial_docs/Cross-terminal_and_Developer_Ecology/python)
diff --git a/docs/assets/scenario/en/Data_Video.md b/docs/assets/scenario/en/Data_Video.md
deleted file mode 100644
index 949a455e..00000000
--- a/docs/assets/scenario/en/Data_Video.md
+++ /dev/null
@@ -1,14 +0,0 @@
-# Data Video Creation
-Data video is a video format that displays and introduces data content through charts, animations, annotations, voiceovers, etc. It can make complex data more understandable and guide the audience to acquire useful information and insights. Data video is widely used in business presentations, news reports, educational training, and other fields. It is an efficient way of data presentation and is very popular among various short video platforms.
-
-The traditional data video production process requires the creator to draw the data into charts using professional visualization software after collecting the data, or use various chart libraries to code and draw charts and then export them as videos. In the era of short videos, everyone can be the creator of videos. However, non-professional video creators often lack the ability to visualize data into charts and then export them as videos, which greatly increases the threshold for data video production.
-
-Based on VMind and video editing tools, it is easy to create rich and colorful data videos. After generating animated charts using the VMind chart intelligent generation module, users can export them as videos or GIFs, import them into professional video editing software like imovie, premiere pro, and so on, to create a complete data video.
-
-
-
-
-
-By integrating VMind into video editing tools, the creation of data videos will also form a complete cycle.
-
-
diff --git a/docs/assets/scenario/en/Document_Assistant.md b/docs/assets/scenario/en/Document_Assistant.md
deleted file mode 100644
index 2632d786..00000000
--- a/docs/assets/scenario/en/Document_Assistant.md
+++ /dev/null
@@ -1,12 +0,0 @@
-# Document Assistant
-When writing reports or documents, accurate and beautiful charts are indispensable. The Document Assistant built on VMind can help users quickly generate charts from selected data and insert them into documents.
-
-
-
-
-
-
-In addition, VMind will also introduce an intelligent insight feature, which can intelligently identify insights in chart data and automatically add annotations to them. These annotations can directly display specific data values, provide additional explanations and contextual information, and help readers better understand the charts.
-
-
-
diff --git a/docs/assets/scenario/en/OpenManus_VMind.md b/docs/assets/scenario/en/OpenManus_VMind.md
deleted file mode 100644
index fc8c75b9..00000000
--- a/docs/assets/scenario/en/OpenManus_VMind.md
+++ /dev/null
@@ -1,257 +0,0 @@
-# OpenManus X VMind: Quickly Build Your Data Analysis Assistant
-
-## Introduction to OpenManus
-
-
-[OpenManus](https://github.com/FoundationAgents/OpenManus) is a concise and versatile Agent framework. Through the pluggable design of Tools and Prompts, it supports various tool invocation capabilities such as Computer Use, Browser Use, and Planning Use.
-
-In OpenManus, you can use the basic manus agent to try out various tasks and conduct interesting explorations. You can also use OpenManus as a platform to unleash your creativity. By writing unique tools and prompts, you can create a one - of - a - kind agent.
-
-**Raise your hand 🙋, so the question is, are there any examples?**
-
-**Hello, yes, examples are coming right up!**
-
-## DataAnalysis Agent
-
-### Data Analysis Tasks
-Our lives are filled with a vast amount of data in various industries such as economy, tourism, the Internet, and retail. If you ask DeepSeek how much data we can generate in a day, DeepSeek will tell you: The International Data Corporation (IDC) predicts that the global annual data volume will reach **175ZB (zettabytes)** in 2025, which is equivalent to approximately **479EB (exabytes)** per day (1EB = 10^18 bytes).
-
-In such a large amount of data, data analysis, data presentation, and final data consumption have always been hot topics. How to use AI tools to assist in analysis or even let them do the analysis for you is the focus. The remarkable data analysis ability of Manus has left a deep impression on everyone. Turning to OpenManus, can OpenManus do what Manus can?
-
-**The answer is: No, but also yes.**
-
-Perhaps directly using OpenManus cannot achieve the same effect as Manus, because OpenManus is a very concise and versatile Agent framework, and its tool library does not have suitable data analysis tools yet. However, precisely because OpenManus is a very simple and versatile Agent framework, as long as you dare to add tools to the library, OpenManus will use them, ultimately achieving a 1 + 1 > 2 effect.
-
-So, we, [VisActor](https://www.visactor.com/) , in collaboration with the OpenManus team, integrated the open - source intelligent visualization solution VMind to add a DataAnalysis Agent to OpenManus.
-
-### Integration Method
-Compared with the application ideas of general Agents in the data analysis scenario, we have made some minor changes, as follows:
-
-
-
-
-General OpenManus uses general Python tools to generate various Python code tasks. In the data analysis scenario, this includes but is not limited to data processing, data reporting, and data visualization.
-
-Our change is to extract the data visualization part into a separate tool, making the application scenario of this tool more focused and in - depth. The functions of this tool include data visualization, data insight, and adding data insights to charts. For details, please refer to [ReadMe](https://github.com/FoundationAgents/OpenManus/blob/main/app/tool/chart_visualization/README.md):
-
-
-
-The original Python tool will then focus more on data processing/data reporting tasks.
-
-#### Python Tool Modification
-Thanks to the existing manus architecture, adding a dedicated Python tool has become very smooth and simple. Observing the original Python tool, we can find that its structure is also very simple and clear, as follows:
-- The name, description, and parameters directly correspond to the prompt parameters of the function call in the large language model.
-- The execute function is the corresponding function executed when the current Python tool is called, and it will ultimately call the _run_code function to obtain the result.
-
-```python
-class PythonExecute(BaseTool):
- """A tool for executing Python code with timeout and safety restrictions."""
-
- name: str = "python_execute"
- description: str = "Executes Python code string. Note: Only print outputs are visible, function return values are not captured. Use print statements to see results."
- parameters: dict = {
- "type": "object",
- "properties": {
- "code": {
- "type": "string",
- "description": "The Python code to execute.",
- },
- },
- "required": ["code"],
- }
-
- def _run_code(self, code: str, result_dict: dict, safe_globals: dict) -> None
-
-
- async def execute(
- self,
- code: str,
- timeout: int = 5,
- ) -> Dict:
- """
- Executes the provided Python code with a timeout.
-
- Args:
- code (str): The Python code to execute.
- timeout (int): Execution timeout in seconds.
-
- Returns:
- Dict: Contains 'output' with execution output or error message and 'success' status.
- """
-```
-
-What we do is very simple. We only need to inherit this function and modify the existing prompt to make it a tool suitable for data analysis. The result is as follows. We mainly clarify the following points through the prompt:
-
-- In the prompt, clarify that the main tasks are data analysis, data reporting, and other ordinary code writing tasks.
-- Hope the large language model can provide dataset overview, details of each column in the dataset, basic data statistics, generate derived indicators, time - series comparison, outliers, key insights, etc.
-- Guide the large language model to conduct multiple analyses based on existing analyses to form in - depth and meaningful analysis data.
-
-```python
-from app.config import config
-from app.tool.python_execute import PythonExecute
-
-class NormalPythonExecute(PythonExecute):
- """A tool for executing Python code with timeout and safety restrictions."""
-
- name: str = "python_execute"
- description: str = """Execute Python code for in-depth data analysis / data report(task conclusion) / other normal task without direct visualization."""
- parameters: dict = {
- "type": "object",
- "properties": {
- "code_type": {
- "description": "code type, data process / data report / others",
- "type": "string",
- "default": "process",
- "enum": ["process", "report", "others"],
- },
- "code": {
- "type": "string",
- "description": """Python code to execute.
-# Note
-1. The code should generate a comprehensive text-based report containing dataset overview, column details, basic statistics, derived metrics, timeseries comparisons, outliers, and key insights.
-2. Use print() for all outputs so the analysis (including sections like 'Dataset Overview' or 'Preprocessing Results') is clearly visible and save it also
-3. Save any report / processed files / each analysis result in worksapce directory: {directory}
-4. Data reports need to be content-rich, including your overall analysis process and corresponding data visualization.
-5. You can invode this tool step-by-step to do data analysis from summary to in-depth with data report saved also""".format(
- directory=config.workspace_root
- ),
- },
- },
- "required": ["code"],
- }
-
- async def execute(self, code: str, code_type: str | None = None, timeout=5):
- return await super().execute(code, timeout)
-```
-
-Through the above operations, we can obtain a tool proficient in using Python code for data analysis/data reporting by simply modifying the prompt. Adding it to the manus tool library completes your customized creation.
-
-#### Data Analysis Agent
-After having our dedicated tool, it becomes even simpler to form our data analysis agent!
-
-**Step 1: Inherit ToolCallAgent**
-
-ToolCallAgent is a general agent that calls various tools. By inheriting it, we already have the ability to call tools to solve problems.
-
-```python
-""" Inherit ToolCallAgent """
-class DataAnalysis(ToolCallAgent):
- """
- A data analysis agent that uses planning to solve various data analysis tasks.
-
- This agent extends ToolCallAgent with a comprehensive set of tools and capabilities,
- including Data Analysis, Chart Visualization, Data Report.
- """
-```
-
-**Step 2: Rewrite the prompt**
-
-Modify the system_prompt and next_step_prompt to inject the soul of the agent.
-
-```python
-class DataAnalysis(ToolCallAgent):
- """
- A data analysis agent that uses planning to solve various data analysis tasks.
-
- This agent extends ToolCallAgent with a comprehensive set of tools and capabilities,
- including Data Analysis, Chart Visualization, Data Report.
- """
-
- name: str = "DataAnalysis"
- description: str = "An analytical agent that utilizes multiple tools to solve diverse data analysis tasks"
-
- # Modify your own prompt
- system_prompt: str = SYSTEM_PROMPT.format(directory=config.workspace_root)
- next_step_prompt: str = NEXT_STEP_PROMPT
-```
-
-**Step 3: Inject Tools**
-
-The last step is to inject usable tools, and then we're done.
-
-```python
-class DataAnalysis(ToolCallAgent):
- """
- A data analysis agent that uses planning to solve various data analysis tasks.
-
- This agent extends ToolCallAgent with a comprehensive set of tools and capabilities,
- including Data Analysis, Chart Visualization, Data Report.
- """
-
- name: str = "DataAnalysis"
- description: str = "An analytical agent that utilizes multiple tools to solve diverse data analysis tasks"
-
- system_prompt: str = SYSTEM_PROMPT.format(directory=config.workspace_root)
- next_step_prompt: str = NEXT_STEP_PROMPT
-
- max_observe: int = 15000
- max_steps: int = 20
-
- # Inject usable tools
- available_tools: ToolCollection = Field(
- default_factory=lambda: ToolCollection(
- NormalPythonExecute(),
- VisualizationPrepare(),
- DataVisualization(),
- Terminate(),
- )
- )
-```
-### Final Results
-#### Manus
-
-So, what's the final result? Can it reach the level of Manus? Let's find out. We tried a classic analysis case on Manus:
-- [Original case replay address](https://manus.im/share/c3onakN6Iajcm1Vt1xAVG7?replay=1)
-- [Address of all generated products in the case](https://github.com/VisActor/VMind/tree/develop/docs/assets/openManus)
-
-
-
-
-This is a typical data analysis scenario. Given monthly sales data, Manus was expected to provide a plan for a 10% sales increase in the next month. Manus completed this task very well, generating a detailed report and many data analysis results and data visualization results.
-
-#### OpenManus Running Results
-
-After adding VMind to OpenManus, the data analysis and chart generation capabilities of OpenManus have significantly improved on the basis of the original general architecture. It can generate multiple high - quality Markdown data analysis reports and rich and accurate data visualization results.
-
-**Task Decomposition**
-
-The agent will decompose the overall task into sub - tasks such as data processing, data analysis, data visualization, pattern recognition, strategy generation, and final report generation for execution.
-
-
-
-**Final Report**
-
-After all tasks are completed, we generated the final data report and multiple intermediate analysis result reports. The data report also contains links to the generated visualization results.
-
-
-
-**Visualization Results**
-
-The following is a partial display of data visualization results. It can generate interactive chart pages (rendered by VChart) with data annotations.
-
-
-
-## In Conclusion
-### Summary
-
-Thanks to the concise and versatile design of OpenManus, adding a new tool or agent on its basis is a very simple task 🚀 Developers only need to focus on the specific implementation and input - output of their own tools, without spending a lot of energy on the agent framework, interaction, and communication with large language models. They can easily create their own agents! The DataAnalysis Agent is such an example. We only need to design the corresponding data analysis tools to have super - strong data analysis capabilities. Perhaps this is the feeling of standing on the shoulders of giants!
-
-### Next Steps
-
-This is just a small attempt. There are still many problems at present. In the future, we will continue to improve the scheduling ability of data analysis tools and integrate a more powerful VMind mcp service. Our goal is to build a full - process tool integrating data analysis, chart recommendation, automatic beautification, automatic layout, and automatic report generation. The products include charts, reports, posters, dashboards, and presentations.
-
-### Welcome to Communicate and Contribute
-
-Finally, we welcome everyone to participate in the open - source construction of OpenManus and VisActor in various ways, such as submitting issues, writing code, adding comments, and giving stars. These are all forms of support for open - source projects.
-
-OpenManus: https://github.com/FoundationAgents/OpenManus
-
-VMind: https://github.com/VisActor/VMind
-
-VisActor Feishu communication group:
-
-
-
-VisActor WeChat official account:
-
-
\ No newline at end of file
diff --git a/docs/assets/scenario/en/Open_API.md b/docs/assets/scenario/en/Open_API.md
deleted file mode 100644
index fc0db826..00000000
--- a/docs/assets/scenario/en/Open_API.md
+++ /dev/null
@@ -1,165 +0,0 @@
-# VMind Open API
-For some scenarios that are not suitable for accessing the front-end SDK (such as Python Notebook or backend scenarios) or that are inconvenient to apply for LLM service, we also provide Open API service, users can use VMind through HTTP requests to meet diverse chart visualization needs.
-
-## Open API request example:
-```curl
-curl --location 'VMind Open API service address' \
---header 'api-key: service secret key' \
---form 'data="[
-{
-\"Product Name\": \"Cola\",
-\"region\": \"south\",
-\"sales\": 2350
-},
-{
-\"Product Name\": \"Cola\",
-\"region\": \"east\",
-\"sales\": 1027
-},
-{
-\"Product Name\": \"Cola\",
-\"region\": \"west\",
-\"sales\": 1027
-}
-]"' \
---form 'model="doubao"' \
---form 'userPrompt="Show me the sales of products in the north"' \
-```
-
-Return Value:
-```json
-{
-"chart": [
-{
-"chartType": "BAR CHART", // Chart type
-"spec": { // Generated VChart Spec
-"type": "bar",
-"xField": [
-"region",
-"Product Name"
-],
-"yField": "sales",
-"seriesField": "Product Name",
-//...Omit more spec content
-},
-"usage": { // LLM token consumption details
-"completion_tokens": 19,
-"prompt_tokens": 1109,
-"total_tokens": 1128
-},
-"time": { // Chart animation duration
-"totalTime": 2000,
-"frameArr": []
-}
-}
-]
-}
-```
-
-The VMind Open API currently supports multiple models for intelligent data aggregation and chart generation, as well as chart-advisor for rule-based chart recommendations. The specific supported models are as follows:
-- doubao: Default value, corresponding to the doubao-pro-32k-240828 model (Pro model released on 240828)
-- doubao-1.5-pro-32k: Corresponding to the Doubao 1.5 Pro model released on 250115
-- doubao-1.5-lite-32k: Corresponding to the Doubao 1.5 Lite model released on 250115
-- Others: For example, GPT series models and DeepSeek series models
-
-## Interface Description
-### /generateChart
-- Generate vchart spec for chart rendering according to the data and user's input natural language.
-- If userPrompt and model are empty, the LLM will not be called, and the chart recommendation will only be based on the data to generate a series of chart spec, sorted by recommendation degree.
-- If data is empty, a spec template without data will be generated, and the fillSpecWithData interface can be filled with data later.
-- VMind ensures that only data field information (field names, field types) are passed to the LLM, and detailed data will not be passed to the LLM.
-#### Invocation Method
-POST
-#### Parameter Description
-- data
-Description: The dataset used in the chart, with a JSON array structure. It can be empty. If data is empty, a spec template will be generated, and the fillSpecWithData interface can be used to fill in the data later.
-
-- userPrompt
-Description: Optional, user's input in natural language, string type. It can specify chart type and field allocation, and further perform data aggregation, filtering, sorting.
-Examples:
-```
-Show me the sales of each product in different regions
-Show the sales data of each product using a line chart
-Top three products in terms of sales in the north
-Which product sold the most
-Sales status of products in different regions
-Show the sales data of each product, with region as the x-axis and product type as the color
-```
-
-- model
-Optional values: doubao | doubao-1.5-pro-32k | doubao-1.5-lite-32k
-Description: Optional. Specifies the model type, defaults to the doubao model. Also supports GPT series and DeepSeek series models, available upon request.
-
-- mode
-Optional value: "accuracyFirst" ， "performanceFirst"
-Description: Optional, chart generation mode, whether to prioritize performance or accuracy. AccuracyFirst will improve the accuracy of data aggregation and chart generation, but the generation speed will be affected. PerformanceFirst will improve the speed of chart generation and data aggregation, but accuracy may be affected. The default is performanceFirst
-
-- enableDataQuery
-Optional value: true or false
-Description: Optional, whether to enable intelligent data aggregation. After opening, it will support data aggregation, filtering, sorting, grouping, etc. in userPrompt, such as:
-```
-Show me the sales of products in the north
-Show the top 3 products sold in the south
-Generate a line chart, aggregated by date
-Show the top 5 products in terms of sales
-```
-The default is true, enabling it will increase the interface response time.
-
-- chartTypeList
-String array
-Description: Optional, chart type list, used to restrict the chart types generated by the chart generation service. Only chart types in this list will be generated. If this parameter is not passed, all chart types will be generated by default.
-
-Optional values:
-
-```
-"Bar Chart"
-"Line Chart"
-"Pie Chart"
-"Scatter Plot"
-"Word Cloud"
-"Rose Chart"
-"Radar Chart"
-"Sankey Chart"
-"Funnel Chart"
-"Dual Axis Chart"
-"Waterfall Chart"
-"Box Plot"
-"Dynamic Bar Chart"
-```
-
-Example:
-```json
-[
-"Bar Chart",
-"Line Chart",
-"Pie Chart",
-"Scatter Plot"
-]//Only generate bar chart, line chart, pie chart, scatter plot
-```
-
-#### Return Value
-- chart
-
-Description: Chart generation result, json array structure
-
-```typescript
-type chartGenerationResult = {
-chart:{
-chartType : string; //Chart type
-spec: Spec; //The vchart spec corresponding to this chart type, which can be used for front-end rendering
-score: number; //Chart recommendation score, the higher the score, the more suitable the current data is for displaying with this type of chart
-chartSource: string //Chart source, chartAdvisor represents rule-based chart recommendation, does not call the LLM; if the LLM is called to complete the chart generation, it will be the corresponding model name
-}[]
-}
-```
-
-## Contact Information
-If you are interested in VMind Open API integration, please join the Lark group to create topics, or follow our WeChat public account to join the WeChat group and contact the relevant personnel.
-
-- Lark:
-
-
-
-- WeChat Public Account (you can join the WeChat group and Lark group through the public account menu):
-
-
diff --git a/docs/assets/scenario/en/Wordcloud_Cube.md b/docs/assets/scenario/en/Wordcloud_Cube.md
deleted file mode 100644
index abeac622..00000000
--- a/docs/assets/scenario/en/Wordcloud_Cube.md
+++ /dev/null
@@ -1,26 +0,0 @@
-# Visual Narration - Word Cloud Cube
-VisActor has been consistently dedicated to exploring the application of data visualization in narrative scenarios. VMind will also be combined with VChart narrative template to launch the intelligent narrative function.
-
-We have combined VMind with the Coze platform to launch the Word Cloud Cube robot. Users only need to upload the title and core words of the word cloud, and VMind will intelligently generate personalized word clouds of various themes for users.
-
-
-
-
-
-
-
-
-
-Although VMind itself has the ability to automatically infer chart types based on data and intelligently generate species, constraints need to be added to the results under the scenario of Word Cloud Cube. The VMind node service has encapsulated the VMind component, only allowing VMind to choose and generate from our specified word cloud template.
-
-
-
-In the future, we will add a variety of theme templates, such as holidays, zodiacs, etc., to provide more possibilities and inspirations for users' creations. At the same time, we will also support the generation of GIF animations, allowing the word cloud to show a unique dynamic effect, giving it stronger visual impact and expressiveness.
-
-Of course, we will also break through the limitations of word clouds and upgrade the Word Cloud Cube to a "Card Cube" which can create with a wider range and more stunning effects. In the meantime, we will also promote the overall upgrade of VMind to make it more intelligent, achieve qualitative improvement in all aspects such as code generation, chart configuration, automatic beautification, story arrangement, and serve more developers through open source services.
-
-## Experience Link:
-- [Coze](https://www.coze.cn/s/iY7WbPQ2/)
-
-## Related Reading
-- [Full Bloom in One Thought, Coze Combined with VisActor Text to Picture Combat Summary](https://juejin.cn/post/7340307383505436712)
diff --git a/docs/assets/scenario/menu.json b/docs/assets/scenario/menu.json
deleted file mode 100644
index d5b25661..00000000
--- a/docs/assets/scenario/menu.json
+++ /dev/null
@@ -1,47 +0,0 @@
-{
- "menu": "scripts",
- "children": [
- {
- "path": "OpenManus_VMind",
- "title": {
- "zh": "OpenManus X VMind : 快速打造你的数据分析助手",
- "en": "OpenManus X VMind: Quickly Build Your Data Analysis Assistant"
- }
- },
- {
- "path": "Chart_Assistant",
- "title": {
- "zh": "图表助手",
- "en": "Chart Assistant"
- }
- },
- {
- "path": "Wordcloud_Cube",
- "title": {
- "zh": "可视化叙事-词云魔方",
- "en": "Visual Narration: Word Cloud Cube"
- }
- },
- {
- "path": "Data_Video",
- "title": {
- "zh": "数据视频创作",
- "en": "Data video creation"
- }
- },
- {
- "path": "Document_Assistant",
- "title": {
- "zh": "文档助手",
- "en": "Document Assistant"
- }
- },
- {
- "path": "Data_Development",
- "title": {
- "zh": "数据开发",
- "en": "Data Development"
- }
- }
- ]
-}
\ No newline at end of file
diff --git a/docs/assets/scenario/zh/Chart_Assistant.md b/docs/assets/scenario/zh/Chart_Assistant.md
deleted file mode 100644
index 67f1b564..00000000
--- a/docs/assets/scenario/zh/Chart_Assistant.md
+++ /dev/null
@@ -1,15 +0,0 @@
-# 图表智能助手
-VMind可以成为大语言模型的可视化代理(agent)，负责处理图表生成、图表编辑等相关任务，从而打造出“图表智能助手”。这个助手既可以作为一个独立的机器人提供服务，也可以以Copilot形式接入其他产品。
-
-
-
-
-
-
-作为图表智能助手，VMind可以根据用户输入的自然语言描述，自动生成相应的图表。还可通过智能数据聚合模块实现数据的二次筛选、聚合、排序。
-
-无论你需要展示的是时间序列的折线图，还是不同类别的柱状图，VMind都能迅速响应并帮助你选择图表类型及生成图表。有了VMind，用户无需深厚的图表设计知识，就可以制作出既美观又信息丰富的图表，极大地提高了工作效率。
-
-## 体验地址:
-- [Coze](https://www.coze.com/s/Zs8MNnSod/)
-- [Discord](https://discord.gg/sajqrVSP): 加入Discord频道，在聊天中 @VMind Chart Master 机器人即可体验
diff --git a/docs/assets/scenario/zh/Data_Development.md b/docs/assets/scenario/zh/Data_Development.md
deleted file mode 100644
index 9bd9b650..00000000
--- a/docs/assets/scenario/zh/Data_Development.md
+++ /dev/null
@@ -1,57 +0,0 @@
-# 数据开发
-在数据开发等python Notebook场景中，适合在SQL查询结束后，调用VMind Open API进行图表生成，配合py-vchart渲染图表，在Notebook中集成智能可视化能力。
-
-
-
-
-
-## 示例代码
-```python
-'''
-在Notebook场景下，可在SQL查询后调用VMind Open API进行图表生成，并结合py-vchart进行图表渲染
-VMind支持基于大模型的图表生成和基于规则的图表生成
-图表生成过程不会将明细数据传给大模型，尽可放心食用
-'''
-
-import requests
-import pandas as pd
-from pyvchart import render_chart
-
-url = "" #VMind OPen API url
-
-# 查询出的数据集：
-print('sql查询结果：')
-display(sales_data_df)
-
-sales_data_json =sales_data_df.to_json(orient="records", force_ascii=False)
-
-userPrompt = "帮我展示不同区域各商品销售额" # 用户的展示意图
-
-print()
-print('输入展示意图：')
-print(userPrompt)
-print()
-
-body = {
- "data": sales_data_json,
- "userPrompt": userPrompt,
- "model": "doubao" # 使用豆包模型
-}
-
-# 调用VMind open api进行图表生成
-response = requests.post(url, headers=headers, json=body)
-
-chart = response.json().get('chart')[0]
-spec = chart.get('spec')
-chartType = chart.get('chartType')
-
-# print(spec)
-print('生成的图表类型: '+chartType)
-
-# 调用py-vchart渲染图表
-render_chart(spec)
-```
-
-## 相关阅读
-- [VMind Open API](./Open_API)
-- [py-vchart](https://visactor.io/vchart/guide/tutorial_docs/Cross-terminal_and_Developer_Ecology/python)
diff --git a/docs/assets/scenario/zh/Data_Video.md b/docs/assets/scenario/zh/Data_Video.md
deleted file mode 100644
index f7b7333b..00000000
--- a/docs/assets/scenario/zh/Data_Video.md
+++ /dev/null
@@ -1,14 +0,0 @@
-# 数据视频创作
-数据视频是一种通过图表、动画、标注、旁白等方式展示和介绍数据内容的视频形式。它能将复杂的数据变得更加易懂，并引导观众获取有用的信息和见解。数据视频广泛应用于商业展示、新闻报道、教育培训等方面，是一种高效的数据展现方式，受到了各大短视频平台的喜爱。
-
-传统的数据视频制作流程需要创作者在收集数据后，使用专业的可视化软件将数据绘制为图表；或使用各种图表库，通过编码绘制图表并导出为视频。在短视频时代，每个人都可能是视频的创作者。然而，非专业的视频创作者往往不具备将数据可视化为图表并导出为视频的能力，这大大提高了数据视频的制作门槛。
-
-基于VMind 结合视频编辑工具，可以很容易制作丰富多彩的数据视频。使用VMind图表智能生成模块生成带动画的图表后，用户可以将其导出为视频或者GIF，导入到剪映、pr等专业的视频剪辑软件中，制作一个完整的数据视频。
-
-
-
-
-
-将VMind集成到视频剪辑工具中，数据视频的创作也将形成完整的闭环。
-
-
diff --git a/docs/assets/scenario/zh/Document_Assistant.md b/docs/assets/scenario/zh/Document_Assistant.md
deleted file mode 100644
index 7d35dd43..00000000
--- a/docs/assets/scenario/zh/Document_Assistant.md
+++ /dev/null
@@ -1,12 +0,0 @@
-# 文档助手
-在撰写报告或文档时，准确而美观的图表是必不可少的。基于VMind打造文档助手，能够帮助用户快速从选定的数据中生成图表，并插入到文档中。
-
-
-
-
-
-
-此外，VMind还将推出智能洞察功能，这个功能能够智能地识别图表数据中的洞察，并自动为其添加标注。这些标注可以直接展示具体的数据数值，同时提供额外的解释和上下文信息，帮助读者更好地理解图表。
-
-
-
diff --git a/docs/assets/scenario/zh/OpenManus_VMind.md b/docs/assets/scenario/zh/OpenManus_VMind.md
deleted file mode 100644
index 67736be8..00000000
--- a/docs/assets/scenario/zh/OpenManus_VMind.md
+++ /dev/null
@@ -1,249 +0,0 @@
-# OpenManus X VMind : 快速打造你的数据分析助手
-## OpenManus 简介
-
-
-[OpenManus](https://github.com/FoundationAgents/OpenManus) 是一个简洁通用的 Agent 框架，通过可插拔的 Tools 和 Prompt 组合设计，支持 Computer Use、Browser Use 和 Planning Use 等多种工具调用能力
-在OpenManus中，你可以使用基础的manus agent来尝试完成你的各种任务和有趣的探索，也可以以OpenManus为载体，挥洒你自己的创意，通过编写独特的tools和prompt，实现独一无二的agent
-
-**举手🙋，那么问题来了，有比如吗？**
-
-**你好，有的，🌰马上就端上来了！**
-
-## 🌰：DataAnalysis Agent
-
-### 数据分析任务
-
-我们生活中充斥着大量的数据，经济、旅游、互联网、零售等各行各业，如果你去询问DeepSeek一天我们可以产生多少数据，DeepSeek会告诉你：国际数据公司（IDC）预测，2025 年全球年数据量将达 **175ZB（泽字节）**，相当于 **每天约 479EB（艾字节）**（1EB=10^18 字节）
-
-在如此繁多海量的数据中，进行数据分析，数据展示和最终数据消费一直以来是一个热门的话题，而如何使用AI工具辅助分析，甚至帮你分析更是焦点。Manus令人赞叹的数据分析能力给所有人都留下了深刻的印象，转过头看向OpenManus，Manus可以，OpenManus可以吗？
-
-**答案是：不可以，但也可以。**
-
-或许直接使用OpenManus无法达到Manus的效果，因为OpenMaus是个非常简洁通用的Agent框架，其工具库里还没有趁手的数据分析利器；但正因为OpenManus是个非常简单通用的Agent框架，你只要敢往工具库里加，OpenManus就敢用，最后达到1 + 1 > 2的效果。
-
-于是我们 [VisActor](https://www.visactor.com/) 联合OpenManus团队，集成开源智能可视化解决方案 VMind ，为OpenManus 添加一个 DataAnalysis Agent。
-
-### 接入方式
-
-相比于通用的Agent对于数据分析场景的应用思路，我们做了一些些微小的改动，具体如下：
-
-
-
-
-
-通用的OpenManus，会使用通用的python工具去实现各种各样，各种各类的pyhton代码生成任务，在数据分析场景，包括且不限于：数据处理、数据报告、数据可视化等；
-而我们的改动在于，将数据可视化这一部分提取成了一个单独的tool，使得这个工具对应的使用场景更加聚焦和深入，这个tool的功能包括：数据可视化、数据洞察以及将数据洞察添加到图表中三个功能，详细可见[ReadMe](https://github.com/FoundationAgents/OpenManus/blob/main/app/tool/chart_visualization/README_zh.md)：
-
-
-
-而原本的python tool将会更加专注在数据处理/数据报告的任务上。
-#### Python Tool改造
-得益于已有的manus架构，新增一个专属python tool变的十分的丝滑和简单，观察原有的python tool，可以发现其结构也十分简单清晰，具体如下：
-- name/description/parameters直接对应了大模型中function call的prompt参数。
-- execute是当前python tool被调用时执行的对应函数，而执行中最终会调用_run_code函数得到结果
-
-```python
-class PythonExecute(BaseTool):
- """A tool for executing Python code with timeout and safety restrictions."""
-
- name: str = "python_execute"
- description: str = "Executes Python code string. Note: Only print outputs are visible, function return values are not captured. Use print statements to see results."
- parameters: dict = {
- "type": "object",
- "properties": {
- "code": {
- "type": "string",
- "description": "The Python code to execute.",
- },
- },
- "required": ["code"],
- }
-
- def _run_code(self, code: str, result_dict: dict, safe_globals: dict) -> None
-
-
- async def execute(
- self,
- code: str,
- timeout: int = 5,
- ) -> Dict:
- """
- Executes the provided Python code with a timeout.
-
- Args:
- code (str): The Python code to execute.
- timeout (int): Execution timeout in seconds.
-
- Returns:
- Dict: Contains 'output' with execution output or error message and 'success' status.
- """
-```
-而我们做事情非常简单，我们只需要继承这个函数，改一下已有的prompt，将其引导成适合做数据分析的工具就可以了！结果如下所示，主要通过prompt明确以下几个内容
-- 在prompt中明确任务主要为：数据分析，数据报告和其他普通的代码编写任务
-- 希望大模型可以提供：数据集总览，数据集各列详情，基础的数据统计信息，生成派生指标，时序对比，异常值，关键洞察等
-- 引导大模型可以根据已有的分析多次分析，形成深入有意义的分析数据
-
-```python
-from app.config import config
-from app.tool.python_execute import PythonExecute
-
-class NormalPythonExecute(PythonExecute):
- """A tool for executing Python code with timeout and safety restrictions."""
-
- name: str = "python_execute"
- description: str = """Execute Python code for in-depth data analysis / data report(task conclusion) / other normal task without direct visualization."""
- parameters: dict = {
- "type": "object",
- "properties": {
- "code_type": {
- "description": "code type, data process / data report / others",
- "type": "string",
- "default": "process",
- "enum": ["process", "report", "others"],
- },
- "code": {
- "type": "string",
- "description": """Python code to execute.
-# Note
-1. The code should generate a comprehensive text-based report containing dataset overview, column details, basic statistics, derived metrics, timeseries comparisons, outliers, and key insights.
-2. Use print() for all outputs so the analysis (including sections like 'Dataset Overview' or 'Preprocessing Results') is clearly visible and save it also
-3. Save any report / processed files / each analysis result in worksapce directory: {directory}
-4. Data reports need to be content-rich, including your overall analysis process and corresponding data visualization.
-5. You can invode this tool step-by-step to do data analysis from summary to in-depth with data report saved also""".format(
- directory=config.workspace_root
- ),
- },
- },
- "required": ["code"],
- }
-
- async def execute(self, code: str, code_type: str | None = None, timeout=5):
- return await super().execute(code, timeout)
-```
-
-通过以上操作，我们通过简单的prompt修改，就可以得到一个精通python代码进行数据分析/数据报告的工具，将其加入到manus的使用工具库中，就完成了你的专属打造。
-
-#### 数据分析Agent
-
-在有了我们的专属Tool之后，形成我们的数据分析Agent就更加简单了！
-
-**第一步：继承ToolCallAgent**
-
-ToolCallAgent是一个调用各种tool的通用agent，我们继承它就已经拥有了调用tool解决问题的能力
-```python
-""" 继承ToolCallAgent """
-class DataAnalysis(ToolCallAgent):
- """
- A data analysis agent that uses planning to solve various data analysis tasks.
-
- This agent extends ToolCallAgent with a comprehensive set of tools and capabilities,
- including Data Analysis, Chart Visualization, Data Report.
- """
-```
-
-**第二步：改写prompt**
-
-修改system_prompt和next_step_prompt,注入agent的灵魂
-```python
-class DataAnalysis(ToolCallAgent):
- """
- A data analysis agent that uses planning to solve various data analysis tasks.
-
- This agent extends ToolCallAgent with a comprehensive set of tools and capabilities,
- including Data Analysis, Chart Visualization, Data Report.
- """
-
- name: str = "DataAnalysis"
- description: str = "An analytical agent that utilizes multiple tools to solve diverse data analysis tasks"
-
- # 修改你自己的prompt
- system_prompt: str = SYSTEM_PROMPT.format(directory=config.workspace_root)
- next_step_prompt: str = NEXT_STEP_PROMPT
-```
-
-**第三步：Tool注入**
-
-最后一步，注入可使用的Tool，就大功告成
-```python
-class DataAnalysis(ToolCallAgent):
- """
- A data analysis agent that uses planning to solve various data analysis tasks.
-
- This agent extends ToolCallAgent with a comprehensive set of tools and capabilities,
- including Data Analysis, Chart Visualization, Data Report.
- """
-
- name: str = "DataAnalysis"
- description: str = "An analytical agent that utilizes multiple tools to solve diverse data analysis tasks"
-
- system_prompt: str = SYSTEM_PROMPT.format(directory=config.workspace_root)
- next_step_prompt: str = NEXT_STEP_PROMPT
-
- max_observe: int = 15000
- max_steps: int = 20
-
- # 注入可使用的tools
- available_tools: ToolCollection = Field(
- default_factory=lambda: ToolCollection(
- NormalPythonExecute(),
- VisualizationPrepare(),
- DataVisualization(),
- Terminate(),
- )
- )
-```
-### 最终效果
-#### Manus
-那么最终效果如何呢，是否能够企及Manus的高度呢？让我们一试便知；
-我们尝试Manus上的一个经典分析案例：
-- [原始案例回放地址](https://manus.im/share/c3onakN6Iajcm1Vt1xAVG7?replay=1)
-- [案例中所有生成产物地址](https://github.com/VisActor/VMind/tree/develop/docs/assets/openManus)
-
-
-
-
-这是一个典型的数据分析场景，给Manus一个月度的销售数据，期望给出下个月10%销售增长的计划；Manus很好的完成了这个任务，生成了一个详细的报告，并且产生了很多数据分析的数据结果和数据可视化结果。
-
-#### OpenManus 运行结果
-在OpenManus上加入VMind之后，OpenManus的数据分析能力和图表生成能力在原有的通用架构基础上有了显著的提升，可以生成多个高质量的markdown数据分析报告和丰富准确的数据可视化结果。
-
-**任务拆解**
-
-Agent会将整体任务拆分成：数据处理、数据分析，数据可视化，模式识别，策略生成和最终报告生成这几个子任务进行执行
-
-
-
-**最终报告**
-
-在所有任务完成之后，我们产生了最终的数据报告和多个中间分析结果的报告，数据报告中同样包含生成的可视化结果链接
-
-
-
-
-**可视化结果**
-
-以下是部分数据可视化结果展示，可以生成可交互式的图表页面（VChart渲染）并且带有数据标注
-
-
-
-## 写在最后
-### 总结
-得益于OpenManus简洁通用的设计，在其基础上新增Tool或者Agent是一件相当简单的事情🚀
-开发者只要关注自己工具的具体实现和输入输出，而不用花大量精力在Agent的框架，交互以及与大模型的通信上，轻松打造属于自己的Agent！
-DataAnalysis Agent就是这样的一个例子，我们只需设计对应的数据分析工具，就能拥有超强的数据分析能力，或许，这就是站在巨人肩膀上的感觉吧！
-### 下一步
-这只是我们一个小小的尝试，目前还有很多问题，后续我们会继续完善 数据分析工具的调度能力，以及集成更强大的 VMind mcp 服务。我们的目标是打造一个集数据分析、图表推荐、自动美化、自动布局、自动报告生成的全流程工具。 产物包括图表、报告、海报、看板以及演示文稿。
-### 欢迎交流与共建
-最后，欢迎大家以各种方式参与 OpenManus 和 VisActor 的开源建设，提 issue，写代码，加评论，点 Star，都是对开源项目的支持。
-
-OpenManus：https://github.com/FoundationAgents/OpenManus
-
-VMind：https://github.com/VisActor/VMind
-
-
-VisActor 飞书交流群：
-
-
-VisActor 微信公众号：
-
-
diff --git a/docs/assets/scenario/zh/Open_API.md b/docs/assets/scenario/zh/Open_API.md
deleted file mode 100644
index 143322b3..00000000
--- a/docs/assets/scenario/zh/Open_API.md
+++ /dev/null
@@ -1,165 +0,0 @@
-# VMind Open API
-对于某些不适合接入前端SDK（例如Python Notebook或后端场景）或者不方便申请LLM服务的场景，我们还提供了Open API服务，用户可以通过HTTP请求的形式使用VMind，满足多样化的图表可视化需求。
-
-## Open API请求示例：
-```curl
-curl --location 'VMind Open API服务地址' \
---header 'api-key: 服务密钥' \
---form 'data="[
- {
- \"商品名称\": \"可乐\",
- \"region\": \"south\",
- \"销售额\": 2350
- },
- {
- \"商品名称\": \"可乐\",
- \"region\": \"east\",
- \"销售额\": 1027
- },
- {
- \"商品名称\": \"可乐\",
- \"region\": \"west\",
- \"销售额\": 1027
- }
-]"' \
---form 'model="doubao"' \
---form 'userPrompt="帮我展示北方各商品销售额"' \
-```
-
-返回值：
-```json
-{
- "chart": [
- {
- "chartType": "BAR CHART", //图表类型
- "spec": { //生成的VChart Spec
- "type": "bar",
- "xField": [
- "region",
- "商品名称"
- ],
- "yField": "销售额",
- "seriesField": "商品名称",
- //...省略更多spec内容
- },
- "usage": { //LLM token消耗明细
- "completion_tokens": 19,
- "prompt_tokens": 1109,
- "total_tokens": 1128
- },
- "time": { //图表动画时长
- "totalTime": 2000,
- "frameArr": []
- }
- }
- ]
-}
-```
-
-VMind Open API目前支持多种模型进行智能数据聚合和图表生成，以及chart-advisor完成基于规则的图表推荐,具体支持模型如下所示：
-- doubao: 默认值,对应doubao-pro-32k-240828模型，即240828发布的pro模型
-- doubao-1.5-pro-32k: 对应250115发布的豆包1.5pro模型
-- doubao-1.5-lite-32k: 对应250115发布的豆包1.5lite模型
-- 其他:例如gpt系列模型和deepseek系列模型
-
-## 接口说明
-### /generateChart
-- 根据数据和用户输入的自然语言，生成vchart spec用于图表渲染。
-- 若userPrompt和model为空，则不调用大模型，仅根据数据完成图表推荐，生成一系列图表spec，根据推荐程度排序。
-- 若data为空，则生成不包含数据的spec模板，后期可调用fillSpecWithData接口填充数据。
-- VMind确保仅将数据字段信息（字段名称、字段类型）传递给大模型，明细数据不会传递给大模型。
-#### 调用方法
-POST
-#### 参数说明
-- data
-说明：图表中使用的数据集，json 数组结构。可以为空。data为空则生成spec模板，后期可调用fillSpecWithData接口填充数据。
-
-- userPrompt
-说明：可选，用户输入的自然语言，字符串类型。可指定图表类型和字段分配，可进一步进行数据聚合、筛选、排序。
-示例：
-```
-帮我展示不同区域各商品销售额
-展示各商品销售数据，使用折线图
-北方销售额前三的商品
-哪种商品卖的最多
-不同地区商品销售状况
-展示各商品销售数据，地区做x轴，商品类型做颜色
-```
-
-- model
-可选值：doubao | doubao-1.5-pro-32k | doubao-1.5-lite-32k
-说明：可选，模型类型，默认支持doubao模型；同时也支持gpt系列和deepseek系列，可以按需提供
-
-- mode
-可选值："accuracyFirst" ， "performanceFirst"
-说明：可选，图表生成模式，性能优先还是准确性优先。accuracyFirst会导致数据聚合和图表生成准确性提高，但生成速度将会受影响。performanceFirst会提高图表生成和数据聚合的速度，但准确性可能会受影响。默认为performanceFirst
-
-- enableDataQuery
-可选值：true或false
-说明：可选，是否开启智能数据聚合。开启后将支持userPrompt中的数据聚合、筛选、排序、分组等指令，例如：
-```
-帮我展示北方各商品销售额
-展示南方卖的最多的3种商品
-生成折线图，按照日期聚合
-展示销量前5名的商品
-```
-默认为true，开启后会增加接口响应时间。
-
-- chartTypeList
-字符串数组
-说明：可选，图表类型列表，用于限制图表生成服务生成的图表类型。仅会生成该列表中的图表类型。不传该参数则默认生成所有图表类型。
-
-可选值：
-
-```
-"Bar Chart"
-"Line Chart"
-"Pie Chart"
-"Scatter Plot"
-"Word Cloud"
-"Rose Chart"
-"Radar Chart"
-"Sankey Chart"
-"Funnel Chart"
-"Dual Axis Chart"
-"Waterfall Chart"
-"Box Plot"
-"Dynamic Bar Chart"
-```
-
-示例：
-```json
-[
-"Bar Chart",
-"Line Chart",
-"Pie Chart",
-"Scatter Plot"
-]//仅生成柱状图、折线图、饼图、散点图
-```
-
-#### 返回值
-- chart
-
-说明：图表生成结果，json数组结构
-
-```typescript
-type chartGenerationResult = {
-chart:{
- chartType : string; //图表类型
- spec: Spec; //该图表类型对应的vchart spec，可用于前端渲染
- score: number; //图表推荐得分，分数越高代表当前数据越适合用这种图表展示
- chartSource: string //图表来源，chartAdvisor表示基于规则的图表推荐，不调用大模型；若调用大模型完成了图表生成，则为对应的模型名
- }[]
-}
-```
-
-## 联系方式
-若有VMind Open API接入意向，欢迎加入飞书群创建话题，或关注微信公众号，加入微信群联系相关人员。
-
-- 飞书：
-
-
-
-- 微信公众号（通过公众号菜单可以加入微信群和飞书群）：
-
-
diff --git a/docs/assets/scenario/zh/Wordcloud_Cube.md b/docs/assets/scenario/zh/Wordcloud_Cube.md
deleted file mode 100644
index 412bae55..00000000
--- a/docs/assets/scenario/zh/Wordcloud_Cube.md
+++ /dev/null
@@ -1,26 +0,0 @@
-# 可视化叙事-词云魔方
-VisActor一直致力于探究数据可视化在叙事场景中的应用，VMind也将与VChart叙事模板相结合，推出智能叙事功能。
-
-我们将VMind与Coze平台结合，推出了词云魔方机器人。用户只需要上传词云的标题和核心词，VMind便能智能地为用户生成各种主题的个性化词云。
-
-
-
-
-
-
-
-
-
-VMind 本身具备根据数据自动推断图表类型，并智能生成spec 的能力，但是在词云魔方场景下，需要对结果进行约束。VMind node 服务对VMind 组件进行了一层封装，只允许VMind 在我们规定的词云模版中进行选择生成。
-
-
-
-未来我们将添加各种各样的主题模板，如节日、生肖等，给用户的创作提供更多可能和灵感。同时，我们还将支持生成GIF动图，让词云展现出独特的动态效果，使其具有更强的视觉冲击力和表达力。
-
-当然，我们也会突破词云的限制，将词云魔方升级成为“卡片魔方”，可以生成使用范围更广，效果更惊艳的作品。同时也借此，整体推动VMind的升级，使其更加智能化，在代码生成，图表配置，自动美化，故事编排等各个方面都能有质的提升，并通过开源服务更多的开发者。
-
-## 体验地址:
-- [Coze](https://www.coze.cn/s/iY7WbPQ2/)
-
-## 相关阅读
-- [一念生花，Coze 结合VisActor 文生图实战总结](https://juejin.cn/post/7340307383505436712)
diff --git a/docs/debugging.md b/docs/debugging.md
new file mode 100644
index 00000000..705a6a4f
--- /dev/null
+++ b/docs/debugging.md
@@ -0,0 +1,73 @@
+# Debug Skills
+
+## Recommended: Codex or Claude Code
+
+Use your existing coding agent for everyday development and debugging. It provides the model, tool execution and conversation management; this repository supplies the Skill. Codex and Claude Code are the documented host targets. Other compatible agents may work, but do not have dedicated setup instructions or compatibility guarantees here.
+
+Both hosts load complete Skill directories. Codex uses `.agents/skills/` in a project; Claude Code uses `.claude/skills/`. See the [official Codex instructions](https://learn.chatgpt.com/docs/build-skills) and [Claude Code instructions](https://code.claude.com/docs/en/skills).
+
+The following local setup uses a fresh directory so it does not overwrite an existing Skill installation. Start in this repository's root, with your chosen agent installed and authenticated:
+
+```bash
+debug_workspace="$(mktemp -d)"
+mkdir -p "$debug_workspace/inputs" "$debug_workspace/outputs"
+cp evals/cases/vchart-skill/inputs/sales.csv "$debug_workspace/inputs/"
+```
+
+Choose one host. For **Codex**:
+
+```bash
+mkdir -p "$debug_workspace/.agents/skills"
+cp -R skills/vchart-skill skills/vtable-skill "$debug_workspace/.agents/skills/"
+cd "$debug_workspace"
+codex
+```
+
+For **Claude Code**, instead run this from the repository root after the shared setup:
+
+```bash
+mkdir -p "$debug_workspace/.claude/skills"
+cp -R skills/vchart-skill skills/vtable-skill "$debug_workspace/.claude/skills/"
+cd "$debug_workspace"
+claude
+```
+
+You can also open that workspace in the corresponding app. The scripts need Python 3.9+, and generated HTML needs a browser and public CDN access. The API harness, uv environment and its model configuration are not required for this path.
+
+In Codex CLI, send:
+
+```text
+$vchart-skill Use every row in inputs/sales.csv to build a grouped bar chart comparing sales by region and quarter. Write editable source and runnable HTML under outputs/.
+```
+
+In Claude Code, send the same request with `/vchart-skill` instead of `$vchart-skill`. For a table, select `vtable-skill` and ask for a sortable detail table. Inspect the generated source and HTML, then continue the conversation with changes or errors to investigate.
+
+After editing this repository's Skill, refresh the installed copy before retesting (or create another fresh workspace with these steps). Record the host/model, prompt, inputs and observed result when reporting a problem. Logs and tool access in this route are managed by the host; they do not automatically appear in `evals/runs/`.
+
+Historical generation/browser runs in this migration used Codex. Claude Code's setup above follows its documented Skill interface; a real Claude Code generation run has not yet been recorded for this migration.
+
+## Alternative: a simple model API harness
+
+Use this when you have an API endpoint or want repeatable runs on selected cases. It loads one Skill and provides the bounded tools needed for these tasks. It does not provide an interactive coding-agent application. Only OpenAI-compatible Chat Completions with tool calling and token usage is supported; native protocols are outside the current scope.
+
+Follow [API setup](../evals/README.md#contributor-quickstart) to configure your endpoint, model and key environment variable. Run a built-in case with:
+
+```bash
+uv run python evals/scripts/run.py --case vchart-grouped --browser
+```
+
+Or supply a task without editing the case catalogue:
+
+```bash
+uv run python evals/scripts/run.py \
+ --skill vchart-skill \
+ --prompt-file evals/examples/chart-prompt.md \
+ --input evals/cases/vchart-skill/inputs/sales.csv \
+ --browser --html demo.html
+```
+
+Use `--skill vtable-skill` for tables. Replace the prompt/data paths with your own files. Relative paths are resolved from your shell's current directory. Each `--input` is copied into `inputs/