> ## Documentation Index
> Fetch the complete documentation index at: https://wonder.design/docs/llms.txt
> Use this file to discover all available pages before exploring further.

# Wonder MCP tools

> Every tool Wonder MCP gives your agent, grouped by what it lets the agent do, with what each role and plan allows and the limits on each call.

See every tool Wonder MCP gives your agent, what each one lets it do, and what your role and plan allow. Your client may show these names when it asks you to approve a tool call.

**Before you start** · Any MCP client · Free plan

## How agents use these tools

Every MCP client gets the same tools, and so do chats in the Wonder desktop app. Your agent picks the tools it needs from what you ask: you don't call them yourself.

The tools run on Wonder's servers, so Wonder doesn't need to be open. Your agent only reaches the workspaces and files your account can open. Tools that read never change anything. Tools that change a file need you to be able to edit it, and so do `finish_artboard` and `list_file_move_destinations` (see [What each role and plan allows](#what-each-role-and-plan-allows)).

## Get context and guidance

These tools tell your agent where it is and how Wonder expects design work to be done.

| Tool | What it lets your agent do |
| - | - |
| `get_design_context` | Get what a task needs in one go: the frames on the page you're on, the file's variables and icons, who you are, and Wonder's design basics. |
| `get_basic_info` | Check again where you are: your workspace, the open file and page, what you've selected and the file's pages. |
| `get_design_guidelines` | Load Wonder's design basics, and the rules of one of its [built-in styles](https://wonder.design/docs/design-with-ai.md#ask-for-a-style-by-name). |
| `get_skills` | Load Wonder's step-by-step instructions for one kind of job, such as a new design, an edit, or bringing code onto the canvas. |
| `get_wonder_info` | Answer your questions about Wonder itself. |

## Read designs

These tools only read, so they work for anyone who can open the file.

| Tool | What it lets your agent do |
| - | - |
| `get_artboards` | List the frames on a page with their position, size and main layers. |
| `get_element_tree` | See the outline of a design: its layers, their names and how they nest. |
| `get_element_code` | Read layers or a whole frame as React and Tailwind code, to build the design in your code. |
| `find_elements` | Find layers from a plain description, such as "the sign-up button". |
| `get_additional_context` | Understand where a layer sits: the frame and parent it's in, the layers beside and inside it, and how many layers it holds. |
| `take_screenshot` | Get a picture of a layer or frame, to check how its work looks. |

## Create and edit designs

| Tool | What it lets your agent do |
| - | - |
| `create_artboard` | Add a new frame to a page, built from code. |
| `update_elements` | Change a frame: add, replace, move or restyle its layers. |
| `duplicate_elements` | Copy a layer or frame, once or many times, with changes to each copy. |
| `delete_elements` | Delete layers, or a whole frame. |
| `rename_elements` | Rename layers and frames. |
| `finish_artboard` | Tell you a frame is ready, with a link to open it. The design is already saved as the agent works. |

Before a big change to one of your designs, agents copy it and change the copy, unless you tell them to edit it directly. See [How agents treat your designs](https://wonder.design/docs/design-with-ai.md#how-agents-treat-your-designs).

## Files, folders and pages

| Tool | What it lets your agent do |
| - | - |
| `list_files` | List the files you can open, with their workspace and folder, or search them by name. |
| `open_file` | Work on a file without you opening it in Wonder. |
| `create_file` | Create a new file and get a link to it. |
| `create_folder` | Create a folder in a workspace. |
| `rename_file` | Rename a file. |
| `list_file_move_destinations` | See which of your workspaces can take a file. |
| `move_file` | Send a file to a different workspace, outside any folder there. You must own or administer the file's workspace and be at least an editor in the other one. Its pin and GitHub connection don't come along. |
| `create_page` | Add a page to a file. People looking at the file aren't taken to it. |
| `rename_page` | Rename a page. |
| `move_page` | Change a page's place in the file's page list. |

## Variables and icons

In these tools, your file's [variables](https://wonder.design/docs/variables.md) are called design tokens.

| Tool | What it lets your agent do |
| - | - |
| `get_design_tokens` | Read the file's variables and their values. |
| `create_design_tokens` | Add variables, including ones that point to another variable. |
| `update_design_tokens` | Change a variable's value or name. Every layer that uses it updates. |
| `delete_design_tokens` | Delete variables. Layers that used them keep their current values. |
| `get_icons` | List the file's [icons](https://wonder.design/docs/icons.md), or read an icon's SVG. |
| `create_icons` | Add SVG icons to the file's icon library. |
| `update_icons` | Give icons new names or keywords. |
| `delete_icons` | Remove icons from the library. Layers that used them keep the icon. |

## Images and vectors

Generating images and vectors needs [Pro](https://wonder.design/pricing), with a monthly allowance per person in each workspace. Uploading and looking at images works on every plan. See [Generate images and vectors](https://wonder.design/docs/generate-images-and-vectors.md).

| Tool | What it lets your agent do | Plan |
| - | - | - |
| `generate_image` | Generate an image from a prompt, on the canvas or inside a design. | Pro |
| `generate_svg` | Generate a vector from a prompt. | Pro |
| `vectorize_image` | Turn an image into a vector. | Pro |
| `prepare_image_upload` | Upload an image from your computer to use in a design. | Free |
| `fetch_image_content` | Look at an image. | Free |
| `fetch_image_description` | Describe an image in words, which is quicker than looking at it. | Free |

## What agents can't do

* **Delete files, folders or pages.** Agents can delete layers, frames, variables and icons, but not a whole file, folder or page. Delete those yourself in Wonder.
* **Organize folders.** Agents can create a folder, but not rename or move one, or put a file into one.
* **Edit where you can't.** An agent has your access, never more. It can't change a file you can only view.

## What each role and plan allows

| You | What your agent can do |
| - | - |
| Owner, admin or editor of a Pro workspace | Use every tool. Moving a file out of the workspace needs its owner or an admin. |
| Owner of a Free workspace | Use every tool except generating images and vectors. |
| Member of a Free workspace who isn't its owner | Read only. On Free, only the owner can edit. |
| Viewer | Read only. |
| Signed in with a personal access token that only reads a workspace | Read only, in that workspace. |

On the Free plan, your agents can make 3,000 AI-generated layers a week, across all your Free workspaces. Your week is the 7 days counted from when you created your account. Once your agents reach it, a request they've already started still finishes, and after that `create_artboard`, `update_elements`, `duplicate_elements` and the image and vector tools are refused until your week resets. Every other tool keeps working. Pro has no weekly layer limit.

When a tool is refused, your agent passes on Wonder's message. See [The agent can read but not edit](https://wonder.design/docs/wonder-mcp-troubleshooting.md#the-agent-can-read-but-not-edit) for what each message means.

## Limits on each call

| Tool | Limit |
| - | - |
| `update_elements` | One frame per call. |
| `duplicate_elements` | Up to 50 copies per call. |
| `delete_elements` | Up to 2,000 layers per call, counting the layers inside them. |
| `find_elements` | Up to 10 descriptions per call. |
| `create_design_tokens`, `update_design_tokens`, `delete_design_tokens` | Up to 100 variables per call. |
| `create_icons` | Up to 20 icons per call. |
| `update_icons` | Up to 100 icons per call. |
| `generate_image`, `generate_svg` | One result per call from an MCP client. In chats in the Wonder desktop app, a vector, or an image added as its own frame, comes from each model you picked, up to 4. |
| `prepare_image_upload` | Images up to 100 MB. |
| `create_page` | Up to 25 pages in a file. |

On every plan, each person can make 120 tool calls a minute, with up to 8 running at once, and each call can run for up to 10 minutes. `get_basic_info` has its own allowance, so checking where you are never uses these up. See [Too many tool calls](https://wonder.design/docs/wonder-mcp-troubleshooting.md#too-many-tool-calls).

## If something goes wrong

If your agent can't find these tools, can't change a file or stops at a limit, see [Wonder MCP troubleshooting](https://wonder.design/docs/wonder-mcp-troubleshooting.md).

## Related

* [Wonder MCP](https://wonder.design/docs/wonder-mcp.md): connect your agent to Wonder.
* [Design with AI](https://wonder.design/docs/design-with-ai.md): what to ask your agent, and how it treats your designs.
* [Generate images and vectors](https://wonder.design/docs/generate-images-and-vectors.md): the Pro image and vector tools in detail.
* [Wonder MCP troubleshooting](https://wonder.design/docs/wonder-mcp-troubleshooting.md): fix connection problems and limits.


This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.