mcp

# SVGator MCP Server — Documentation The SVGator MCP server connects AI assistants to [SVGator](https://www.svgator.com), 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](https://modelcontextprotocol.io) over Streamable HTTP - **Authentication:** OAuth 2.1 (sign in with your SVGator account — see [Getting started](getting-started.md)) ## 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. ## Tools The server exposes 22 tools. See the [Tools reference](tools.md) for full parameter and output tables. **Projects** - **`list_projects`** — List the authenticated user's SVGator projects (newest first), with an `editor_url` deep link per row and a total count for pagination. - **`get_project`** — Get a single project summary: title, preview, folder, timestamps, canvas size, background colour, animation timeline, and saved export settings. - **`create_project`** — Create a new SVGator project from a full project JSON, owned by the authenticated user. - **`replace_project`** — Replace an existing project's entire JSON with a supplied full project (validation runs before any mutation). - **`duplicate_project`** — Make a copy of an existing project, optionally into a different folder or under a new title. - **`move_project`** — Move a project to a different folder and/or rename it. **Inspecting & editing a project** - **`get_skeleton`** — Get a cheap, token-light structure map of a project (element tree with names and keyframe counts only, never values) — step 1 of the editing loop. - **`read_part`** — Read the exact value or element at an id-anchored path — step 2 of the editing loop. - **`edit_part`** — Apply one validated edit (update, insert, move, or delete) at a path and persist it — step 3 of the editing loop. - **`describe_shape`** — Get the exact JSON shape of one building block (element or animator) before authoring, so field names are never guessed. **Assets** - **`list_assets`** — List the authenticated user's reusable assets (static or animated SVG items), newest first. - **`get_asset`** — Get a single asset including its full `data` ({document, definitions}) payload. - **`create_asset`** — Create a new reusable asset from an asset document. - **`update_asset`** — Replace an existing asset's entire document. **Folders** - **`list_folders`** — List the authenticated user's folders (the labels projects are filed under). - **`get_folder`** — Get a single folder (id, title, timestamps) by its id. - **`create_folder`** — Create a new folder to organize projects. - **`update_folder`** — Rename an existing folder. **Export & account** - **`export_project`** — Export a project to a deliverable format (SVG, Lottie, React Native, Flutter, video, or GIF), with optional per-export overrides. - **`get_render`** — Poll the status of a raster (video/GIF) render started by `export_project` and get the signed download URL when done. - **`get_profile`** — Get the authenticated customer profile: name, email, plan, and teams. ## Documentation map | Document | For | Covers | | --- | --- | --- | | [Getting started](getting-started.md) | Everyone | Connecting from claude.ai, Claude Desktop, Claude Code, and other MCP clients; authentication; privacy | | [Tools reference](tools.md) | Integrators / the curious | Every tool with full parameter and output tables | | [Editing projects](editing.md) | Integrators / power users | The navigate → read → edit → verify loop, path addressing, authoring rules | | [Exporting projects](exporting.md) | Everyone | Formats, delivery modes, export options, render polling | | [Skill & resources](skill-and-resources.md) | Integrators | The motion-design craft skill, MCP resources, and the prompt | | [Troubleshooting](troubleshooting.md) | Everyone | Connection issues, common errors, and what they mean | ## Requirements - A **SVGator account** ([sign up](https://app.svgator.com)). 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.