SVGator MCP

official

Create, edit, and export SVGator animations (SVG, Lottie, GIF, video) from your AI assistant, and keep an editable project you can open in the editor anytime.

GitHub

What can you do with Svgator MCP?

Browse and list your projects — ask the AI to list all your SVGator projects with list_projects or get details and the editor_url for a specific one via get_project.
Inspect project structure and animations — use get_skeleton and read_part to see which elements exist in a project and identify their animation properties.
Edit projects incrementally — modify specific elements, such as changing a color or adjusting a loop, using edit_part without rewriting the entire project.
Create or replace entire projects — generate a new animated SVG from scratch with create_project or overwrite an existing one using replace_project.
Manage reusable assets and folders — organize work by creating, updating, and moving assets (create_asset, update_asset) and folders (create_folder, update_folder).
Export projects to deliverable formats — render a project as an MP4 video, Lottie file, or other format using export_project and check the result with get_render.

Documentation

SVGator MCP Server — Documentation

The SVGator MCP server connects AI assistants to SVGator, the online SVG animation tool. Once connected, an assistant can browse, inspect, edit, create, and export your SVGator projects on your behalf — from "list my projects" to "build me an animated logo reveal and export it as a Lottie file".

Endpoint: https://mcp.svgator.com/mcp
Protocol: Model Context Protocol over Streamable HTTP
Authentication: OAuth 2.1 (sign in with your SVGator account — see Getting started)

What you can do

Capability	Tools	Example prompt
Browse your projects	`list_projects`, `get_project`, `get_profile`	"List my SVGator projects"
Inspect a project's structure	`get_skeleton`, `read_part`	"What elements are in project X, and which are animated?"
Edit a project incrementally	`edit_part`	"Make the circle red and loop its bounce forever"
Create or replace whole projects	`create_project`, `replace_project`	"Build me a new animated loading spinner"
Duplicate, move & rename projects	`duplicate_project`, `move_project`	"Copy project X into my 'Drafts' folder"
Manage reusable assets	`list_assets`, `get_asset`, `create_asset`, `update_asset`	"Save this icon as a reusable asset"
Organize projects into folders	`list_folders`, `get_folder`, `create_folder`, `update_folder`	"Make a 'Marketing' folder and file this project under it"
Look up authoring reference	`describe_shape`	(used by the assistant to get exact JSON shapes)
Export to deliverable formats	`export_project`, `get_render`	"Export project X as an MP4 video"
Get motion-design craft guidance	skill resources + prompt	"Art-direct a logo reveal, then build it"

Every project the server returns includes an editor_url — a deep link that opens the project in the SVGator editor, so you can always continue by hand where the assistant left off.

Documentation map

Document	For	Covers
Getting started	Everyone	Connecting from claude.ai, Claude Desktop, Claude Code, and other MCP clients; authentication; privacy
Tools reference	Integrators / the curious	Every tool with full parameter and output tables
Editing projects	Integrators / power users	The navigate → read → edit → verify loop, path addressing, authoring rules
Exporting projects	Everyone	Formats, delivery modes, export options, render polling
Skill & resources	Integrators	The motion-design craft skill, MCP resources, and the prompt
Troubleshooting	Everyone	Connection issues, common errors, and what they mean

Requirements

A SVGator account (sign up). Exports count against your plan's limits exactly as they do in the editor.
To connect from claude.ai / Claude Desktop, custom connectors require a Claude Pro, Max, Team, or Enterprise plan. Claude Code and other MCP clients connect directly to the endpoint.

How it works (one paragraph)

The server is a remote MCP server: your MCP client (Claude, or any client speaking Streamable HTTP) connects to https://mcp.svgator.com/mcp and signs you in through SVGator's regular OAuth login. Your SVGator credentials stay server-side only — the AI model never sees your access token or secret key; it only receives the tool results (project lists, structure maps, exported files). Tool inputs are validated against published schemas, and invalid requests return descriptive errors the assistant can read and correct.