Skip to content
saymock
Product Agents Use cases Developers FAQ Pricing
EN, English
  • Français
  • English
  • Deutsch
  • Español
  • Português
  • Italiano
  • Nederlands
  • 日本語
  • 한국어
  • 简体中文
Theme
Sign in Try it

Agents · Claude Code

Design mockups with Claude Code, on a board your team can edit

Updated November 18, 2026

Ask Claude Code for a screen and it draws it on a Saymock board: real HTML + Tailwind, editable by hand, reviewed by your team in the browser. It can read your codebase to design screens true to your product, handle the comments people leave on the board, then turn the approved screens into code in your project.

Create a free account Set it up

What Claude Code gets

  • A board instead of an image. Each screen is a frame on a multiplayer board, written in HTML with Tailwind classes, a format Claude already writes well. You move, restyle and retype things yourself, no prompt needed.
  • Eyes on its own work. get_screenshot renders a screen, check_layout reports overflows, misalignments and missing icons, and compare_to_reference scores a screen against a screenshot you give it and names what differs.
  • Your team’s comments become tasks. Someone comments on a button, mentions the agent, and Claude makes the change, checks it and answers in the thread.
  • Code out. export_code gives React, Vue or plain HTML with the same Tailwind classes; the plugin’s implement-design skill rebuilds screens with your project’s own components and conventions.
  • Alternatives in place. Ask for three directions and they stay attached to the screen, where people preview them and pick one.

Set it up

  1. Install the plugin, in a terminal:
    claude plugin marketplace add saymock/claude-plugin && claude plugin install saymock@saymock
  2. In Claude Code, run /mcp, choose saymock and Authenticate. Saymock opens in the browser: sign in (or create a free account) and approve.
  3. Start a new session: the tools load at start.

The plugin brings the MCP server and four skills: saymock-design (new screens and flows), design-from-code (your app’s existing screens, from the codebase), implement-design (screens to code) and design-review (UX, consistency, accessibility).

Rather have the server alone, or using the Claude Code extension in VS Code?

claude mcp add --scope user --transport http saymock https://app.saymock.com/mcp

In the VS Code extension, which does not put claude on the PATH: type /mcp in the Claude Code panel and add an HTTP server named saymock, scope user, URL https://app.saymock.com/mcp (Claude Code 2.1.261 or later). If you added the server by hand before installing the plugin, remove it (claude mcp remove saymock) or the tools show up twice.

Things to ask

Use Saymock to design the onboarding of this app: three screens for iPhone, using the colors and fonts in our Tailwind config.

Put our current settings page into Saymock from the code, then propose two cleaner layouts.

Implement the “Checkout” screen from Saymock in this project, with our existing components.

Review with your team

Share the board (a link to view, comment or edit, no account needed to view), then run /saymock:session in Claude Code. Claude waits for comments on the board, makes each change as it is written, checks it and replies in the thread. The session stops by itself after a while without requests, or when someone presses End on the board.

Questions

Does it work with a Claude subscription?

Yes. Saymock never calls a model itself: Claude Code does the work with your own account, Pro, Max or API. Saymock only stores and renders the board.

How much does it cost?

Saymock bills human editors, never agents or their tokens. Free covers 2 human editors per workspace with unlimited agents. See the pricing.

Why not ask Claude Code for HTML files directly?

You can, but you get files nobody else can comment on or tweak without a prompt. On a board, your team reviews in the browser, edits by hand, picks between alternatives and keeps the history, and Claude sees and answers their comments.

How does it compare with Figma’s MCP server?

On the same task, same agent, our benchmark measured equivalent quality with 31% fewer tokens, 37% lower cost and 1.8× the speed. Protocol and every run: the benchmark; the differences in approach: Saymock vs Figma MCP.

The tools do not show up.

Start a new session: Claude Code loads MCP tools at start. /mcp shows the server’s state and signs in again.

Other agents: Cursor, Codex, ChatGPT, Claude on the web and desktop, all of them. What to ask for: screens from your code, design to code, all use cases. A whole project, step by step: designing an app with Claude Code.

See also

  • Cursor
  • Codex
  • ChatGPT
  • Claude
  • VS Code
  • Gemini CLI
  • Le Chat
saymock
Product Agents Developers FAQ Pricing Benchmark Figma MCP alternative Connect your agent Use cases Gallery Guides MCP tools Changelog About Contact Sign in

© 2026 Saymock

Legal notice Terms of use Terms of sale Privacy