maketa-mcp
Click on "Deploy Server".
Wait a few minutes for the server to deploy. Once ready, it will show a "Started" state.
In the chat, type
@followed by the MCP server name and your instructions, e.g., "@maketa-mcpCreate a login screen with an email field, a password field, and a submit button."
That's it! The server will respond to your query, and you can continue using it as needed.
Here is a step-by-step guide with screenshots.
Maketa MCP Server
Give your AI assistant hands on real app-screen mockups.
Maketa is a browser-based screen builder for mobile apps and landing pages. Every
board is stored as an explicit JSON scene model (maketa.board.v2) — so an assistant
does not guess pixels from a screenshot, it reads and writes the same structured objects
a human edits in the editor.
Website: https://maketa.pro
Docs for agents: https://maketa.pro/llms.txt · OpenAPI 3.1: https://maketa.pro/openapi.json
Remote MCP endpoint:
https://maketa.pro/maketa/api/mcp(Streamable HTTP)
Quick start — remote server (recommended, nothing to install)
Claude Code
claude mcp add --transport http maketa https://maketa.pro/maketa/api/mcpYou will be asked to sign in with your Maketa account (OAuth 2.1, PKCE, dynamic client registration — no API key to copy around).
Any MCP client that supports remote servers
{
"mcpServers": {
"maketa": {
"type": "http",
"url": "https://maketa.pro/maketa/api/mcp"
}
}
}Headless (CI, autonomous agents — no browser, no human)
Send a bearer token instead of doing OAuth:
# 1. An agent can issue itself a 14-day sandbox key, no account needed:
curl -X POST https://maketa.pro/maketa/api/agent/register \
-H 'content-type: application/json' \
-d '{"name":"my-agent","purpose":"prototype screens"}'
# → {"key":"mk_sb_…","expiresAt":…,"quotas":{"boards":5,"savesPerDay":200}}
# 2. Call MCP with it:
curl -X POST https://maketa.pro/maketa/api/mcp \
-H "Authorization: Bearer mk_sb_…" \
-H 'content-type: application/json' \
-d '{"jsonrpc":"2.0","id":1,"method":"tools/list"}'A project key (mk_pk_…, issued by a board owner) works the same way and is
permanent, scoped to one board — that is what CI uses to push screens from code.
Credential | Scope | Lifetime |
OAuth access token | all boards of the signed-in user | 1 hour (refresh token: 30 days) |
| one board | until revoked |
| sandbox boards created with that key | 14 days, 5 boards, 200 saves/day |
Related MCP server: Figma Copilot
Quick start — local stdio server
{
"mcpServers": {
"maketa": {
"command": "npx",
"args": ["-y", "maketa-mcp"],
"env": {
"MAKETA_BOARD": "<board id>",
"MAKETA_EDIT": "<edit token>",
"MAKETA_PKEY": "mk_pk_…"
}
}
}
}MAKETA_BOARD and MAKETA_EDIT come from the editor's Share dialog: the edit link
is https://maketa.pro/app/?b=<BOARD>&e=<EDIT>.
Tools
Tool | What it does |
| List the boards (mockup files) the caller can access. |
| List screens of a board with a summary: name, device, size, node count, branches. |
| Read one screen: slots and the node tree. |
| Read the subtree of one node — cheaper than the whole screen on large boards. |
| Create a screen (device preset, size, background). |
| Add a node to a slot or inside another node. |
| Patch a node: size, layout, look, content, action. |
| Move a node to another parent or position in the tree. |
| Delete a node with its subtree. |
| Branch a screen into an editable copy — the only way to redesign a screen that came from code. |
| Upsert a screen from source code by a stable key; every push is a new version ("commit from code"). |
| List the branches of a screen. |
| Get the parent screen of a branch (navigate back to the code original). |
The scene model (maketa.board.v2)
A screen is a tree of nodes, not a list of shapes — the same model a human edits in the browser. Children have no coordinates: position is decided by the parent, exactly like on the three native platforms ("constraints down, sizes up, parent positions").
screen
├─ slots.top one node — header, kept while scrolling
├─ slots.body one node — the scrolling content
├─ slots.bottom one node — tab bar
└─ slots.over[] nodes drawn above everything (sheets, dialogs)Node: { id, type, name, children[], w, h, layout, look, content, action }.
Field | Meaning |
|
|
|
|
|
|
| background, border, radius, shadow, image fill |
| text, placeholder, items, |
| what a tap does: open a screen ( |
Board-level shared holds the header and tab bar shown on every screen; a screen can opt
out with noShared. Layout is computed by the same engine in the browser, in Node and on
the server, so what the assistant builds is what the human sees.
Code ↔ design branching
Screens pushed from source code (kind: "code") are locked. An assistant that wants
to redesign one calls maketa_branch_screen and edits the branch — a proposal that lives
next to the original. When the code moves on and a new version is pushed, existing
branches are flagged baseStale, so nobody silently reviews an outdated design.
Example prompts
"Show me the screens of my Maketa board and describe the checkout one."
"Add a primary button at the bottom of the login screen and link it to the home screen."
"Branch the profile screen that came from code and propose a cleaner layout."
Privacy & security
The server sees only your boards (screen objects), your account e-mail and board ids.
An OAuth token grants read/write to that user's boards only; a project key is limited to
one board; a sandbox key is limited to its own sandbox boards, expires in 14 days and
cannot use paid AI features or billing. Access can be revoked from the Maketa account
page, project keys via POST /project/key/revoke.
Privacy policy: https://maketa.pro/privacy/
Terms: https://maketa.pro/oferta/
License / contact
Maketa is free to use. The MCP server is made by demda.pro, a custom software development studio. Questions: support@maketa.pro
This server cannot be deployed
Maintenance
Related MCP Connectors
- MaketaOAuthpro.maketa
Build and edit app screen mockups and clickable prototypes from your AI assistant.
1 Agent-Native design tool - create and edit visual designs with agent assistance
UI design from prompts, screenshots, and URLs for AI coding agents and theme tokens.
Provides UX capabilities to enhance the design output and understanding of AI systems.
Related MCP Servers
- AlicenseNot gradedqualityDmaintenanceEnables AI assistants to read and modify Figma designs programmatically, supporting design analysis, element creation, text replacement, annotations, auto-layout configuration, and prototype visualization through natural language commands.312 npmMIT
- AlicenseBqualityDmaintenanceEnables AI assistants to interact with Figma designs programmatically, supporting comprehensive design automation including text manipulation with formatting preservation, batch operations, prototyping, auto-layout configuration, and component management.757 npm5MIT
- AlicenseNot gradedqualityCmaintenanceEnables AI assistants to inspect, modify, export, and validate design documents with 47 tools covering design, code generation, branding, and print/mockup workflows.1 npm1MIT
- AlicenseNot gradedqualityBmaintenanceEnables AI agents to read and manipulate visual UI designs drawn on a canvas, translating them into precise JSON specs for production code generation.11,972 npm4ISC