メニュー
Generate downloadable SVG and HTML technical diagrams from natural language or structured data, including system architecture, AI Agent architecture, data flow, flowchart, comparison matrix, sequence, state, and ER diagrams.
デフォルトでは、最初にソースを確認する Prompt が選択されています。直接コマンドに切り替えるか、ローカルコピーをダウンロードすることもできます。
インストールを決める前に、SKILL.md と SkillsMP に表示されている付属ファイルをお読みください。
Codex または Claude でインストール この Prompt をコピーして Codex、Claude、または他のアシスタントに貼り付けると、Skill ページを確認してインストールできます。
直接コマンドでは確認用 Prompt が省略されます。実行前にソースを確認してください。
npx skills add https://github.com/zgiai/zgi --skill architecture-diagram-generatorコマンドは1行のまま表示されます。コピー前に横へスクロールして全体を確認してください。
ローカルで確認しますか？SkillsMP が現在取得できるファイルをダウンロードできます。
Navigate the visible console to a safe internal page when the user asks to open, switch to, or go to another module.
Generate downloadable files from provided content, including basic exports, SVG files, styled DOCX documents, HTML-rendered PDFs, and static editable PPTX decks.
Call workflows explicitly bound to the current Agent as structured process tools.
SOC 職業分類に基づく
SKILL.md を表示中
| name | architecture-diagram-generator |
| description | Generate downloadable SVG and HTML technical diagrams from natural language or structured data, including system architecture, AI Agent architecture, data flow, flowchart, comparison matrix, sequence, state, and ER diagrams. |
| when_to_use | Use this skill when the user asks to create, generate, export, draw, visualize, or produce a technical architecture diagram, system diagram, microservice diagram, frontend/backend diagram, database architecture diagram, API gateway diagram, AI Agent architecture, Agent workflow, LLM workflow, RAG workflow, tool-calling workflow, memory workflow, data flow diagram, process flowchart, business process diagram, approval flow, task execution flow, user operation flow, product comparison, competitor comparison, solution comparison, feature matrix, sequence diagram, state diagram, ER diagram, database design diagram, or architecture visualization for documentation, reports, presentations, or embedding. For casual, vague, incomplete, or non-structured diagram requests, first route through prompt-professionalizer to optimize the diagram prompt and extract structured requirements, then call this skill. |
| provider_type | builtin |
| provider_id | architecture_diagram_generator |
| runtime_type | tool |
| tools | ["generate_architecture_diagram"] |
| max_calls_per_turn | 5 |
| timeout_seconds | 10 |
| display | {"icon":"network","category":"content_creation","scenarios":["content_creation","technical_development"],"label":{"en_US":"Architecture Diagram Generator","zh_Hans":"架构图生成器"},"description":{"en_US":"Designed for system architecture, process, and data-relationship planning; turns natural language or structured data into SVG and HTML architecture, flow, sequence, state, or ER diagrams.","zh_Hans":"适用于设计系统架构、业务流程或数据关系，可将自然语言或结构化数据生成 SVG 和 HTML 架构图、流程图、时序图、状态图或 ER 图。"},"when_to_use":{"en_US":"Use when the answer should include generated technical diagram artifacts.","zh_Hans":"当回答需要生成技术架构图文件时使用。"},"tags":{"en_US":["Architecture","Diagram","Visualization"],"zh_Hans":["架构图","技术图","可视化"]}} |
Use this skill to generate downloadable SVG and HTML technical diagram artifacts. Supported diagram types are system_architecture, agent_architecture, data_flow, flowchart, comparison_matrix, sequence, state, and er.
system_architecture: system architecture, microservice architecture, frontend/backend architecture, database architecture, API gateway architecture, deployment, and component diagrams.agent_architecture: AI Agent architecture, LLM inference flow, tool calling, memory management, multi-agent workflow, and RAG pipeline diagrams.data_flow: data movement from input through processing, storage, and output.flowchart: business process, approval flow, task execution flow, and user operation flow.comparison_matrix: product comparison, competitor analysis, solution selection, and feature matrices.sequence: service interactions, API call order, request/response lifecycle, and interface documentation.state: state machines, lifecycle transitions, and status flow.er: entity relationship diagrams for database design and development communication.diagram_type: system_architecture, agent_architecture, data_flow, flowchart, comparison_matrix, sequence, state, or er.prompt-professionalizer to produce an optimized architecture diagram prompt and structured requirements.request_user_input before calling generate_architecture_diagram.generate_architecture_diagram.groups and each node's group when components belong to modules such as frontend, backend, data, tools, memory, model, output, external systems, infrastructure, or business domains. Use layer only for layout order.call_skill_tool with tool_name set to generate_architecture_diagram.When any required decision is missing or ambiguous, call request_user_input instead of writing a plain clarification message. This ensures the UI renders the clarification as a structured confirmation card with optional quick replies.
Generic architecture or diagram requests are incomplete even when the domain can be guessed. Do not infer system_architecture just because the user says "architecture", do not infer flowchart just because the user describes steps, do not infer data_flow just because data is mentioned, and do not infer sequence just because services communicate. If the diagram type is not explicit, ask.
For generic diagram requests, diagram type, diagram title, scope or boundary, and rendering style are required decisions. Ask for them before reading a diagram reference or calling generate_architecture_diagram.
Use a brief message explaining what needs confirmation, then provide 1-4 focused questions. Include options only when each option is a concrete answer that can be used directly. Omit options for open-ended questions such as the diagram title or system boundary.
After calling request_user_input, stop the turn and wait for the user's answer. Do not call generate_architecture_diagram in the same turn.
Ask about:
simple, business, technical, presentation, paper.left_to_right, top_to_bottom.Example request_user_input payload:
{
"message": "I can generate the technical diagram, but need to confirm a few details first.",
"questions": [
{
"id": "diagram_type",
"question": "Which diagram type should I generate?",
"options": [
{ "label": "system_architecture" },
{ "label": "agent_architecture" },
{ "label": "data_flow" },
{ "label": "flowchart" },
{ "label": "sequence" }
]
},
{
"id": "title",
"question": "What title should be shown on the diagram?"
},
{
"id": "style",
"question": "Which rendering style should I use?",
"options": [
{ "label": "simple" },
{ "label": "business" },
{ "label": "technical" },
{ "label": "presentation" },
{ "label": "paper" }
]
},
{
"id": "scope",
"question": "What system boundary or key components must be included?"
}
]
}
Read exactly one reference after choosing the diagram type:
| Requested diagram | Read reference |
|---|---|
system_architecture, microservices, frontend/backend, API gateway, database architecture | diagram-system-architecture.md |
agent_architecture, AI Agent, LLM, RAG, tool calling, memory workflow | diagram-agent-architecture.md |
data_flow, data pipeline, input-process-store-output chain | diagram-data-flow.md |
flowchart, business process, approval flow, task flow, user operation flow | diagram-flowchart.md |
comparison_matrix, feature matrix, product comparison, solution selection | diagram-comparison-matrix.md |
sequence, API calls, service interaction, request lifecycle | diagram-sequence.md |
state, state machine, lifecycle transitions | diagram-state.md |
er, ERD, entity relationship, database design | diagram-er.md |
If the user requests a diagram type that is not listed, say it is not supported yet and offer to structure it as the closest supported type only after confirmation.
generate_architecture_diagram accepts:
diagram_type: required. Supported values: system_architecture, agent_architecture, data_flow, flowchart, comparison_matrix, sequence, state, er.title: optional diagram title.description: optional subtitle or source summary.output_filename: optional ASCII filename without extension. Defaults to architecture-diagram.data: required diagram-specific data object.groups and node group fields when there are clear modules. Use layer only for layout order.options: optional rendering options. Common keys: formats, width, height, style, direction, show_legend, show_labels.lifecycle: optional file lifecycle, persistent or temporary. Defaults to persistent.Default output formats are svg and html.
technical for engineering documentation and architecture reviews.business for stakeholder reports and product comparisons.presentation for slide-ready diagrams.paper for warm, report-ready Paper.design-like visuals.simple when no visual style is specified.generate_architecture_diagram, use prompt-professionalizer when the user's request is casual, vague, incomplete, or not already structured for a technical diagram. Direct tool calls are allowed only when the diagram type, content, and key rendering requirements are already complete.generate_architecture_diagram until the selected diagram reference has been read.request_user_input.