Skip to content

ascii-art

Use when creating ASCII art, CLI banners, text-based illustrations, or decorating command-line tool output with visual elements.

ModelSource
sonnetpack: creative
Full Reference

┏━ 🔧 ascii-art ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┓ ┃ Use when creating ASCII art, CLI banners, text… ┃ ┗━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┛

Create high-quality ASCII art for standalone use or CLI integration. This skill handles discovery (understanding what the user wants), builds a structured brief, and dispatches to the ascii-art-creator agent for production. Understand before drawing — build a brief, dispatch an agent, iterate on results.

AspectValue
Character rangeASCII 32-126 by default; Unicode only when brief permits
Aspect ratio2 cols = 1 visual unit width (2:1 compensation required)
Size tiersSmall 16-30 cols, Medium 40-60 cols, Large 60-80 cols
Max CLI width80 columns hard limit
EmojiNever — breaks monospace alignment
StylesSilhouette, Filled, Mixed, Scene layering
AgentDispatch to ascii-art-creator — never produce art directly
I want to…File
Choose characters for density, texture, line art, box drawing, or full density palettereference/characters.md
Plan sizing, aspect ratio compensation, and dimensions before drawingreference/proportions.md
Apply layering, depth, shading styles, transition zones, and atmospheric elementsreference/composition.md
Draw animals, text/logos, faces, borders, or landscapesreference/subjects.md
Embed art in a CLI tool — banner headers, error art, string escaping, Unicode safetyreference/cli-integration.md
Run pre-commit QA on a piecereference/quality-checklist.md

Usage: Read the reference file matching your current task from the index above. Each file is self-contained with code examples and inline gotchas.

Clear request (specific subject like “ASCII armadillo” or “CLI banner for Neptune”):

  • Ask ONE quick question to nail size and style: “Want this small and minimal, or large and detailed?”
  • If CLI context is obvious, note it in the brief — no need to ask
  • Build brief and dispatch

Ambiguous request (e.g. “make some cool ASCII art”):

  • Ask targeted questions ONE AT A TIME: subject, purpose, size, style/mood
  • Stop asking once you have enough — don’t interrogate
  • Infer reasonable defaults from short answers and note assumptions

Before dispatching, assemble a brief with these fields:

Subject, Size (small/medium/large or specific dimensions), Style (silhouette/filled/mixed/scene), Context (standalone/CLI banner/help screen/error art), Constraints (max width, no Unicode, specific escaping, language), Notes (additional context from discovery).

CLI detection: If the user mentions a CLI tool, command-line app, terminal output, or code integration, set Context to the appropriate CLI variant and include the target language in Constraints automatically.

Dispatch the ascii-art-creator agent using the Task tool. Include the filled-in brief, instruction to read the reference files before starting, and any specific user preferences discovered.

When the user wants changes: identify what to change, update only relevant brief fields, re-dispatch with the updated brief AND the previous art for targeted edits. Only start from scratch if the user requests it, proportions are fundamentally wrong, or the style needs a complete change.

MistakeFix
Skipping discovery, guessing what user wantsAlways ask at least one clarifying question
Asking too many questionsClear requests need ONE question max
Not detecting CLI contextIf user mentions a tool/app/terminal, set CLI context automatically
Dumping art knowledge into the dispatchThe agent loads reference files — just send the brief
Starting from scratch on every iterationSend previous art with refinement notes for targeted edits
Producing art yourself instead of dispatchingYou are the orchestrator. The agent produces art. Always dispatch.