Language:JapaneseEnglish

canvas-design Claude Code Skill

Author
Anthropic
Topics
Design & Media Production · Documents & Content Creation
License
See LICENSE.txt
First cataloged
2026-07-09 (UTC)
Explanation last updated
2026-08-21 (UTC)
Source (GitHub) last updated
2026-04-20 (UTC) (129 days ago)

The explanation below is AI-generated. Please verify it against the sources.

This skill guides Claude through creating original visual artwork, such as posters or design pieces, delivered as .png or .pdf files. It works in two stages: first drafting a written design philosophy (an aesthetic concept described in a markdown file), then translating that philosophy into an actual visual composition. The instructions stress that the final artifact should be mostly visual, with only minimal, purposeful text, and that the work should avoid resembling generated or template-based design. It also directs Claude to avoid copying any existing artist's work, keeping outputs original. An optional multi-page mode allows producing a set of related but distinct visual pages bundled into one file.

What you can do with canvas-design

  • Request a poster, art piece, or static visual design and receive it as a .png or .pdf file
  • Get an accompanying markdown file describing the underlying design concept/aesthetic direction
  • Ask for a single polished visual composition with minimal, deliberate text placement
  • Request multiple related pages/variations bundled into one output for a cohesive visual set
  • Rely on the skill's built-in instruction to avoid replicating any specific existing artist's style or work

Sources

Back to list