360 Secure SkillHub
360 安全技能中心

scholarplotskill

该 Skill 通过 scholarplot-sci-figure MCP 服务器调用 ScholarPlot 服务,依据自然语言描述生成或编辑 SCI 期刊风格图形,如折线图、柱状图、热力图、神经网络架构图和实验流程图,可指定图形类型、风格与数据值,返回高分辨率或矢量图片及元数据。需先配置 MCP 服务与 API Key,实际生成在外部服务完成。

peizhou
peizhou
热度 18.58
最近更新 2026-06-03 23:23:06
分类:创作依赖:MCP 服务 / 第三方 API应用场景:图像生成 / 研究 / 概念解释

安装技能

复制下面的话术发给你的 Agent,它会通过 secure-skills 检索并安装该技能。

执行命令安装此 skill: npx secure-skills-cli add scholarplotskill
Metadata
name
scholarPlotSkill
description
ScholarPlot AI Academic Figure Generator. Connects via MCP to generate and edit SCI-standard figures, including line charts, bar charts, heatmaps, neural network architectures, and experimental flowcharts. Triggers on requests like "generate loss curve", "draw neural network", "create experimental flowchart", "edit figure". Requires API key from figure.thirdme.com.

scholarPlotSkill

ScholarPlot is an AI academic figure generation tool designed for researchers. It uses an MCP (Model Context Protocol) server to instantly generate professional, SCI-journal standard figures from natural language descriptions.

Core Directives

  • MCP Server Dependency: This skill requires the scholarplot-sci-figure MCP server to be configured and running with a valid API key from https://figure.thirdme.com/mcp.
  • Intelligent Generation: Use the generate_sci_figure MCP tool to create figures. Always pass a detailed description. You can optionally specify figureType, style (e.g., nature, science, ieee), and exact dataValues for data charts.
  • Interactive Editing: Use the edit_figure MCP tool to modify existing figures. Pass the figureId and specific editInstructions (e.g., "Change the bar color to blue").
  • High-Quality Output: Return the generated figureUrl to the user so they can view and download the high-resolution (300dpi) or vector (SVG) format image. Present the metadata cleanly.

Execution Pipeline

Step 1: Tool Selection

Determine if the user wants to generate a new figure or edit an existing one based on the conversation context.

Step 2: Invoke MCP Tool

  • For New Figures: Call generate_sci_figure with the provided descriptions and any available data.
  • For Edits: Call edit_figure with the target figureId and instructions.

Step 3: Present Results

Once the MCP tool returns a response, format the output nicely for the user:

  • Display the figure using Markdown image syntax: ![Generated Figure](figureUrl) (if supported by the client) or provide the direct URL.
  • Summarize the metadata (e.g., Type, Style, Resolution).
  • Provide the figureCode if the user wants to review the underlying representation.

Configuration Requirements

Users must configure their Claude Desktop (or other MCP-compatible client) with the following in mcp.json or claude_desktop_config.json:

{
  "mcpServers": {
    "scholarplot-sci-figure": {
      "url": "https://figure.thirdme.com/api/mcp-sse?key=YOUR_API_KEY"
    }
  }
}