graphicslisted

# /graphics — Canonical Graphics Skill This is the ONE skill that handles every graphic the user generates. Eight curated styles (A through H), one decision tree, one example brief the user opens to pick by looking, one safe pipeline. It replaces and consolidates: `infographic`, `visual-assets`, `image-pipeline`, `notebooklm-infographics`, `notebooklm-style-infographics`, `ai-image-diagram`, `image-generation`, `gpt-image-2`, `nano-banana-course-banners`, `whiteboard-explainer-slide`. Every prompt template, parameter, recipe, and safety guarantee from those skills lives inside this skill (see `styles/` and `prompts/`). --- ## When invoked — open the example brief FIRST When `/graphics` is invoked, the first action is: ```bash open ~/.claude/skills/graphics/example-brief.html ``` the user picks a style by looking at examples — A through H — then types or selects their answer in the brief and clicks **Copy My Answers**. The agent reads their answer and proceeds with the matched style recipe in `styles/<letter>-<name>.md`. If the user describes the image without opening the brief, propose a default style based on the description, but ASK FIRST (below) before generating. ### the project Special Case For the project projects, the canonical output is HTML template files following the `example-brief.html` structure, not generated images. Use the exact prompt templates from `output/graphics/hermes-vs-claude-whiteboard-canonical.html` and related files in the the project out