Agent Skills › madhvantyagi/Gnos › excalidraw

excalidraw

GitHub

基于 Excalidraw MCP 服务器创建教学示意图,用于清晰展示空间关系、流程或状态。指导如何加载规范、构建场景并检查图表准确性,确保术语和视觉元素与课程内容一致。

skills/excalidraw/SKILL.md madhvantyagi/Gnos

Trigger Scenarios

需要绘制表示连接、包含、状态变化或边界跨越的空间关系图时 课程需要清晰的流程、状态或边界草图以辅助学习理解时

Install

npx skills add madhvantyagi/Gnos --skill excalidraw -g -y
More Options

Use without installing

npx skills use madhvantyagi/Gnos@excalidraw

指定 Agent (Claude Code)

npx skills add madhvantyagi/Gnos --skill excalidraw -a claude-code -g -y

安装 repo 全部 skill

npx skills add madhvantyagi/Gnos --all -g -y

预览 repo 内 skill

npx skills add madhvantyagi/Gnos --list

SKILL.md

Frontmatter
{
    "name": "excalidraw",
    "description": "Create inspectable teaching diagrams with the Excalidraw MCP server when a lesson needs a clear relationship, process, state, or boundary sketch."
}

Excalidraw diagrams

Use Excalidraw when the learner needs to inspect a spatial relationship: what connects, contains, changes state, or crosses a boundary. The selected subject guide decides what must be true in the picture. This guide decides how to draw and check it. A diagram should answer one question, not decorate a lesson.

Load only the instructions you need

  1. Read this page before using the excalidraw MCP server.
  2. Read color and type for every diagram. Set the meaning of each color and the label hierarchy before drawing.
  3. Read the relevant pattern in diagram types for the relation in the lesson brief. Combine patterns only when the claim requires it.
  4. Read review and handoff before sending the result to the lesson coordinator.

The link to this page loads guidance only. A selected Excalidraw block is produced by calling the excalidraw MCP server's read_me and create_view tools, then inspecting the returned scene.

Keep the diagram's terms, symbols, units, arrow direction, and color meanings identical to the surrounding explanation and exercises. If the brief does not specify these, establish them from the active lesson.

Build the scene

  1. Write the claim in one sentence: “A learner can see that …”. List the objects and the relation needed to support that claim.
  2. Call read_me before the first create_view in a task, unless its current element format is already in context. Its schema is authoritative; do not guess field names, fonts, export commands, or link format.
  3. Sketch the reading order and choose a camera that makes labels readable at the inline viewport. Place boundaries behind the objects they contain.
  4. Call create_view with the schema returned by read_me. Draw in the order the learner should encounter the idea. Keep a final overview in frame.
  5. Open the rendered view and perform the checks in review and handoff. Revise a specific defect; stop when the relation is accurate and legible.

If the server fails, make one correction based on its error. If it remains unavailable, use prose, a table, Mermaid, or a runnable trace and tell the coordinator what was substituted. Do not claim an Excalidraw view or ready artifact without an inspected result.

Version History

  • c0bef66 Current 2026-09-27 10:19

Same Skill Collection

skills/course-design/SKILL.md
skills/course-viewer/SKILL.md
skills/jsxgraph/SKILL.md
skills/khan-academy/SKILL.md
skills/learner-tracking/SKILL.md
skills/learning-orchestrator/SKILL.md
skills/learning/SKILL.md
skills/lesson-design/SKILL.md
skills/manim-voice-animation/SKILL.md
skills/pdf/SKILL.md
skills/pinepaper/SKILL.md
skills/subject/SKILL.md

Metadata

Files
0
Version
a1c2b4d
Hash
a91291ee
Indexed
2026-09-27 10:19

Home - Wiki
Copyright © 2011-2026 iteam. Current version is 2.155.2. UTC+08:00, 2026-10-18 20:11
浙ICP备14020137号-1