--- name: using-craft description: Use when starting any conversation, before responding to any request that involves designing, reviewing or changing a UI, a screen, a page, a flow, a component or a style guide. --- If you were dispatched as a subagent to do one specific task, ignore this skill and do the task. If there is even a small chance that a design skill applies to what you are about to do, invoke it **before you respond**, including before clarifying questions, before reading files and before drawing anything. A design skill that applies is not optional. Small edits are the only exception (see below). ## How 1. Check the routing table below. The one thing to look at first is whether `DESIGN.md` exists at the project root; it decides between `craft-setup` and `craft`. Don't guess it. 2. Invoke the skill with your Skill tool, and announce: "Using to ." 3. Follow it exactly, including its stops for the person's approval. A stop means asking the question and waiting for the answer: through your agent's question tool when it has one (see below), otherwise by ending your turn with the question, which is always possible, even in a headless or one-shot session. Never decide a stop for the person. ## Asking the person When your agent has a tool that asks the person a question with choices, craft's questions go through it: interview questions, stops and picks between options. Choices in a dialog are quicker to answer than options read in chat. - Interview questions go one per call, the way the skills say one per message. Two stops may share a call, as they may share a message. - The options the skill offers become the tool's choices, your recommendation first. Leave out "something else" when the tool has its own free-text answer. More options than the tool takes: keep the ones that fit this project best. - A question with nothing to choose from (which products they like) is asked in plain text. Invented choices steer the answer. - The question stands on its own: name what it is about, because the person may see only the dialog. - Something long to approve (a `DESIGN.md` draft, a states list) is shown in your message first. The tool only asks for the yes. Without such a tool, or when the call fails (as in a headless session), ask in plain text and end your turn. ## Routing table | The situation | Skill | |---|---| | Any request to design, create, mock up or redesign a page, screen, flow or element | `craft` | | A design task in a project with **no `DESIGN.md`** at its root | `craft-setup` first (it offers to import a design system already in code), then `craft` | | "Set up the design", "define our style / brand / guidelines" | `craft-setup` | | "Options", "alternatives", "other directions", "I don't like it", "rethink this" | `craft-explore` | | "Review", "refine", "polish", "make it better", "feels off", or you are about to show your own design | `craft-critique` | | Tokens, the style guide, a new or changed component, "make consistent", "audit" | `craft-system` | | "Import / bring our design system, tokens or components into "; "apply / push / sync the guide's changes to the code" | `craft-sync` | | "What states are missing?", "is this ready for dev?" | `craft-states` | A reusable piece for the system ("we need a date picker component") goes to `craft-system`; something on a specific page goes to `craft`. These are craft's skills (prefixed `craft:` in the Claude Code plugin); prefer them over same-named skills from elsewhere. ## Priority 1. `craft-setup` before `craft` when there is no `DESIGN.md`. Reviews, options for an existing screen, audits and state questions don't need a brief: they work from what's already there. 2. `craft` before its sub-skills: it calls `craft-states`, `craft-explore`, `craft-system`, `craft-sync` and `craft-critique` at the right moment, so don't start them yourself in the middle of a `craft` run. 3. `craft-critique` is the last step before you show anything you made. ## Small edits: no process A change the person described exactly (one value, one label, one element's size, one colour, a typo) is made directly. No skill, no states list, no options. - If the thing is shared (a button style used on many screens), change it **once where it is defined**, and say which screens it affects. Don't open a full design pass for it. - If you're unsure whether it's small, it isn't. Use the table. ## Red flags | Thought | Reality | |---|---| | "It's just a quick mockup, I'll skip the process" | A quick mockup is still a design. The states list takes a minute. Use `craft`. | | "There's no `DESIGN.md`, but the request is clear enough" | Without a brief, every screen invents its own style. `craft-setup` first; the person can say "skip". | | "I'll draw it now and review it later" | Showing unreviewed work trains the person to do your review. `craft-critique` runs before you show. | | "I know which states it needs" | You know the happy path. `craft-states` lists the rest. | | "Let me read the files first, then decide on a skill" | Skills tell you what to read. Decide first. | | "This is a small change" (to a new page) | A new page is never a small change. | | "The session can't wait for answers, so I'll decide and list my calls" | Ask the question: with your agent's question tool, or by ending your turn with it. Deciding for them is the one thing the stops exist to prevent. | ## Other agents Where this skill isn't loaded automatically, put the routing table above in your `AGENTS.md` (or the equivalent) so the agent sees it at the start of every session.