Prompts / start_here
Start here: what do you want to do? Free
The router. Asks what you have and what you want, then points you to the right tool or prompt. Same as getting_started.
Run it
- Claude Code:
/mcp__designers__start_here - Claude Desktop and claude.ai: plus menu, Add from Designers, Start here: what do you want to do?
- Any app: Use the start_here prompt from the Designers server on what you have, or what you want to do.
Arguments
| Name | Meaning |
|---|---|
have | What you have, or what you want to do |
* required
What your assistant is told to do
The full instructions, so you can see exactly what you are asking for:
The person is new to the Designers MCP server. Explain it in plain language and help them start. Do not assume they know what MCP, a tool or a prompt is.
What they have: <have>
1. In three short sentences say what this server is: a set of checks and expert instructions that let you (the AI assistant) review design work properly. It does the exact things models are bad at (colour contrast maths, reading a live page for accessibility problems, spotting generic AI-made design and writing) and gives you the method that a senior design lead would use for critiques, heuristic reviews, research and handoff.
2. Call whoami to show their tier and how many calls they have left today, in one line.
3. Suggest exactly three first tasks that fit what they have. Pick from:
- Figma link: an accessibility review of the frame (prompt accessibility_review), a critique for the stage the work is in (prompt critique), a heuristic evaluation of a flow (prompt heuristic_evaluation).
- Live URL: accessibility check (audit_page_accessibility), design slop check (detect_design_slop), heuristic review (prompt heuristic_evaluation).
- Copy or UI strings: microcopy review (review_ui_copy, prompt microcopy_review).
- Research idea or draft questions: prompt research_plan, then lint_research_questions on the guide.
- Event list or analytics plan: lint_tracking_plan, or prompt tracking_plan for a new product.
- Nothing yet: ask what they are designing and at what stage, then use plan_critique to plan a review.
4. For each task, give the one-line request they can paste next, with their own details filled in where you know them.
5. Mention in one sentence that some deeper prompts, site-wide crawls and scored reports need Pro ($5 a month, 7-day trial) and which free task covers the same ground at smaller scale. Do not push it.
Keep the whole reply under 250 words. End by asking which of the three they want to start with.