Monocolor Editorial Print
Turn a theme, sentence, or photo into an original one-ink or controlled two-ink editorial print image for posters, zines, and packaging.
Evidence shows a static skill package with no executable code or network calls, so no permission requests, data flow, or external side effects. However, the repository provides no documentation on permissions, data handling, or user confirmation mechanisms. Therefore, all trust criteria score 0.
Self-consistency: README, SKILL.md, and design-system catalogs are consistent in visual rules and defaults, but no actual run validation is provided. Dependency availability: Only Python standard library and GitHub Actions are used, but specific versions are not listed. Failure messages: Documentation mentions returning a prompt when image generation is unavailable, but no error handling details are provided.
Audience and scenarios: Multiple use cases are listed, but target users are not defined. Capability boundaries: Clearly states what it does not do in 'Not this' section. Trigger precision: Trigger rules are provided in SKILL.md, but not detailed. Environment fit: Installation to Claude Code skills directory is described, but compatibility is not mentioned.
Information architecture: Repository structure is clear with README, SKILL.md, design-system, etc. Install notes: Provides git clone command. Naming stability: Version 1.2.0, but no naming convention. Examples and FAQ: Multiple examples provided, but no FAQ. Known limitations: Listed in 'Not this', but not explicitly. License: MIT license, but example images have additional copyright. Versioning changelog: CHANGELOG.md exists, but content not provided. Maintenance responsibility: Author info clear, but no maintenance commitment.
Output usability: Describes output format, but no actual output examples. Marginal value: Provides unique visual system, but no comparison with other tools. Cost benefit: No performance or cost data.
Claim traceability: README claims validation scripts, but no results provided. Cross-source corroboration: Only relies on internal repository files. Fact-inference separation: Documentation mixes descriptions and inferences without clear distinction.
- The repository provides no documentation on permissions, data flow, or user confirmation, but as a static skill package, risk is low.
- Example images have separate copyright from the MIT license; be aware of usage restrictions.
- Validation scripts are claimed but no actual results are provided; static review cannot confirm their effectiveness.
What does this agent do, and when should you use it?
This repository is a Claude Code skill that transforms a theme, sentence, object, or supplied photograph into an original editorial-style print image. It defaults to controlled two-ink printing (70%-85% dominant plate, 15%-30% accent plate), with pure one-ink mode available. The visual system emphasizes halftone, risograph grain, cyanotype exposure, or photocopy breakup effects, clear paper visibility with 25%-55% empty space, and type with tension and range. The skill includes a comprehensive design-system catalog (design-system/) and validation scripts to enforce rules and originality. After installation, users trigger it with natural-language prompts, and the generated images are original creations rather than copies of references.
The skill reads color, type, layout, and rhythm rules from the design-system/ catalog, parses user input (theme, phrase, or photo), determines the composition and plate roles, assigns ink (from a choice of one-ink or two-ink recipes), and then composes a prompt for generating a raster image. It also outputs a short recipe naming the print mode, exact ink hex codes, layout family, type pairing, and originality changes. The skill ships with examples and validation scripts (scripts/build_design_system_board.py, scripts/validate_evals.py, scripts/validate_design_system.py); GitHub Actions runs these on every push. It produces a final image when available, and if image generation is unavailable, it outputs only the production-ready prompt and states the limitation.
- Event posters: for parties, exhibitions, city walks, and conceptual posters.
- Social media covers: eye-catching covers for Xiaohongshu, WeChat articles, podcasts, and editorial illustrations.
- Brand collateral: postcards, invitations, tickets, menus, and packaging stickers.
- Keepsakes: travel journals, photo album covers, and anniversary cards.
- Books and publications: covers, title pages, chapter openers, and zine interiors.
- Literary excerpts: illustrations for poetry, literary quotes, or personal declarations.
What are this agent's strengths and limitations?
- Rigorous visual rules: clear limits on ink count (max two), paper visibility, and empty space, emphasizing originality over copying references.
- Customizable: built-in one-ink and two-ink palettes, print processes (halftone, risograph, etc.), and typography system adapt to different content.
- Built-in validation and evaluation: design-system catalog and scripts enforce consistency and quality.
- Versatile use cases: suitable for posters, social media, packaging, and more.
- Platform lock-in: it is a Claude Code skill, requiring the Claude Code environment.
- Setup overhead: users must clone into a specific directory and restart Claude Code.
- Dependency on image generation tools: if unavailable, only a prompt is returned.
- Narrow style scope: focuses on monochrome/two-color printing, not suitable for full-color or complex illustrations.
- Learning curve: users need to understand unique terminology (e.g., 'dominant plate', 'accent plate').
How do you install or deploy this agent?
Clone the repository into your Claude Code skills directory:
git clone https://github.com/yanliudesign/mono-color-skill.git ~/.claude/skills/mono-colorRestart Claude Code after installation to activate the skill.
How do you use this agent?
After installation, restart Claude Code and invoke with natural language, e.g.,
“Use mono-color to make a vertical poster about a midnight convenience store. The exact headline is ‘still open’.”
The skill will output the generated image (if available) along with the production-ready prompt and recipe. Other agent environments can load SKILL.md as the skill entry point.