| name | diagram |
| description | Turn an architecture, workflow, or system description into a source-backed diagram artifact for Codex users. |
Diagram
Use this skill when the user asks to make, draw, visualize, or refine a diagram.
This port is adapted from garrytan/gstack at commit 11de390be1be6849eb9a15f91ff4922dd16c589a.
Workflow
- Clarify the diagram target only if the request is ambiguous:
- audience
- format
- scope
- level of detail
- Choose the simplest useful representation:
- Mermaid for architecture, flowcharts, sequences, and dependency maps
- Markdown tables for matrices and responsibility maps
- generated or editable image artifacts only when the user asks for a visual deliverable
- Build from real repo or source evidence when the diagram describes an existing system.
- Keep the diagram source editable and include it in the answer or in a requested artifact file.
- Verify syntax when a local renderer or parser is available.
Codex Adaptation
- Prefer inline Mermaid for quick diagrams because Codex can render it directly in chat.
- Use local files only when the user asks for a saved artifact or the diagram is too large for chat.
- Do not rely on upstream Claude paths such as
~/.claude/skills/gstack or bundled Excalidraw helpers.
- If image generation or conversion tools are unavailable, return the editable source and state the missing renderer plainly.
Guardrails
- Do not invent system structure when repo evidence is available.
- Do not produce a decorative diagram when the user needs an operational one.
- Do not hide uncertainty; mark inferred edges or speculative components.
- Keep labels short enough to render cleanly.
