| name | ascii-diagrams |
| description | Draw compact text diagrams for code and docs. |
| when_to_use | Use when the user asks to draw or clarify structure, flow, layout, topology, state, memory, or tables in plain text, code comments, READMEs, ADRs, or technical docs. |
ASCII Diagrams
Create compact text diagrams that survive diffs and map to real code or prose.
Overview
- Use a diagram only when it explains structure or flow faster than linear text.
- Keep the skill focused on plain text diagrams, not full diagramming languages.
- Treat
references/ as the pattern library and SKILL.md as the runtime spec.
Principles
- Pick the closest pattern from
references/ before drafting.
- Reuse real identifiers from the user prompt, surrounding code, or nearby docs.
- Show the minimum structure needed to answer the request.
- Keep alignment stable in monospace text.
- Make a minimal best-effort draft when details are missing.
- State assumptions briefly only when they affect interpretation.
Output Contract
Respond in this order:
- Optional one-line title.
- Diagram in a fenced
text block.
1-3 short bullets explaining how to read it.
Do not add extra prose before or after that structure unless the user asks for more.
ASCII vs Unicode
- Default to plain ASCII.
- Use ASCII characters such as
-, |, +, >, <, /, and \ unless a better fit is required.
- Use Unicode only when the user explicitly asks for it, or the destination is clearly rendered Markdown/README and Unicode improves readability.
- Do not use decorative Unicode.
Comment Formatting
- In code comments, preserve the host comment prefix on every line.
- Keep alignment valid after adding
#, //, *, or similar prefixes.
- Prefer ASCII in code comments unless the user explicitly asks for Unicode.
Exclusions
- Do not output Mermaid, PlantUML, or SVG unless explicitly requested.
- Do not output decorative Unicode unless explicitly requested.
- Do not turn the response into generic ASCII art.
References
Match the request to the closest pattern in references/: flowcharts, state machines, trees, data structures, network topology, sequences/tables, or graphs/annotations.