MCP Server AntV
Official# MCP Server AntV  [](https://www.npmjs.com/package/@antv/mcp-server-antv) [](https://smithery.ai/server/@antvis/mcp-server-antv) [](https://www.npmjs.com/package/@antv/mcp-server-antv)
> A **Model Context Protocol (MCP)** server designed for AI development and QA that provides **AntV** documentation context and code examples using the latest APIs.
<img width="768" alt="mcp-server-antv Technical Architecture" src="https://mdn.alipayobjects.com/huamei_qa8qxu/afts/img/A*WHSOR7L8U0YAAAAATjAAAAgAemJ7AQ/fmt.webp" />
Supports **G2**, **G6**, and **F2** libraries for declarative visualization workflows, with **S2**, **X6**, **L7**, and more (including **AVA**, **ADC**, and **G**) coming soon.
## ✨ Features
- ✅ **AntV 5.x Compatibility**: Leverages the latest APIs for performance and modularity.
- 🧩 **Multi-Library Support**: G2 (2D charts), G6 (graph/networks), and F2 (mobile charts).
- 🔍 **Smart Intent Extraction**: Detects library usage and task complexity via `extract_antv_topic`.
- 📚 **Contextual Documentation**: Fetches relevant AntV docs and code snippets with `query_antv_document`.
## 🛠️ Quick Start
### Requirements
- Node.js >= v18.0.0
- Cursor, VSCode, Cline, Claude Desktop or another MCP Client.
### Connect to Cursor
[](https://cursor.com/install-mcp?name=mcp-server-antv&config=eyJjb21tYW5kIjoibnB4IC15IEBhbnR2L21jcC1zZXJ2ZXItYW50diJ9)
Go to: `Settings` -> `Cursor Settings` -> `MCP` -> `Add new global MCP server`
```json
{
"mcpServers": {
"mcp-server-antv": {
"command": "npx",
"args": ["-y", "@antv/mcp-server-antv"]
}
}
}
```
On Window system:
```json
{
"mcpServers": {
"mcp-server-antv": {
"command": "cmd",
"args": ["/c", "npx", "-y", "@antv/mcp-server-antv"]
}
}
}
```
### Connect to VSCode
[](https://insiders.vscode.dev/redirect?url=vscode%3Amcp%2Finstall%3F%257B%2522name%22%3A%22mcp-server-antv%22%2C%22command%22%3A%22npx%22%2C%22args%22%3A%5B%22-y%22%2C%22%40antv%2Fmcp-server-antv%22%5D%7D)
Pasting the following configuration into your VSCode `~/.vscode/mcp.json` file is the recommended approach.
```json
{
"servers": {
"mcp-server-antv": {
"command": "npx",
"args": ["-y", "@antv/mcp-server-antv"]
}
}
}
```
or command-line configuration
```bash
code --add-mcp "{\"name\":\"mcp-server-antv\",\"command\": \"npx\",\"args\": [\"-y\",\"@antv/mcp-server-antv\"]}"
```
## 🧪 Example Workflow
An example workflow:
<video src="https://github.com/user-attachments/assets/242471c6-33f5-4588-95f5-96a3f26aab21" ></video>
## 🧰 Tools Overview
| Tool | Functionality |
| --------------------- | ---------------------------------------------------------------------------- |
| `extract_antv_topic` | Extract user intent, detects library (G2/G6/F2), and infers task complexity. |
| `query_antv_document` | fetch latest documentation and code examples with context7 |
## 🔨 Contributing
Clone the repo
```bash
git clone https://github.com/antvis/mcp-server-chart.git
cd mcp-server-chart
```
Install dependencies:
```bash
npm install
```
Build the server:
```bash
npm run build
```
Start the MCP server:
```bash
npm run start
```
## 📄 License
MIT@[AntV](https://github.com/antvis).
TDQS
Scored across 2 tools
The two tools have clearly distinct purposes: extract_antv_topic handles preprocessing, requirement parsing, and intent extraction, while query_antv_document focuses on retrieving documentation and solutions. Their mandatory sequencing (extract first, then query) reinforces this separation, eliminating any ambiguity about when to use each tool.
Both tools follow a consistent verb_noun pattern with snake_case naming: extract_antv_topic and query_antv_document. The naming clearly indicates their functions (extract vs. query) while maintaining domain specificity (antv_topic vs. antv_document), creating a predictable and readable convention throughout the set.
With only 2 tools, the server feels thin for covering the broad AntV visualization domain (9 libraries mentioned). While the tools logically separate preprocessing from retrieval, many visualization tasks might require additional operations like code generation, validation, or specific library interactions that aren't represented in this minimal set.
The tool surface is severely incomplete for the AntV visualization domain. While the tools cover preprocessing and documentation retrieval, there are significant gaps: no tools for actual code generation, chart rendering, data transformation, style modification, or error handling. Agents will hit dead ends when trying to implement solutions beyond documentation lookup.