Templated MCP Server
OfficialSummary: This MCP server lets AI assistants generate images, videos, and PDFs from Templated templates, and manage templates, folders, uploads, fonts, and account usage.
Render Operations:
create_render(JPG, PNG, WebP, PDF, MP4 with layer modifications, sizing, transparency, animation),get_render,list_renders,delete_render,merge_renders(combine PDFs)Template Operations:
list_templates,get_template,get_template_layers,get_template_pages,create_template,update_template,clone_template,delete_template,list_template_rendersFolder Operations:
list_folders,create_folder,update_folder,delete_folderAsset Operations:
list_uploads,create_upload,delete_upload,list_fonts,upload_font,delete_fontAccount:
get_account(render usage for the current period)
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., "@Templated MCP Server Create an image from my Instagram Post template with the headline 'Big Sale'"
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.
Templated MCP Server
MCP (Model Context Protocol) server for Templated - the API for automated image, video, and PDF generation.
This server enables AI assistants like Claude, Cursor, and ChatGPT to interact with your Templated account - generate renders, manage templates, upload assets, and access all API features using natural language.
Features
Render Generation: Create images (JPG, PNG, WebP), videos (MP4), and PDFs from templates
Template Management: List, create, update, clone, and delete templates
Layer Inspection: Get template layers to understand what can be customized
Asset Management: Upload and manage images, videos, and custom fonts
Folder Organization: Create and manage folders for templates
Related MCP server: HeyGen MCP Server
Quick Start
Get Your API Key
Sign up at app.templated.io
Go to your API Key page
Copy your API key
Connection Options
Option 1: Remote Server (Recommended)
Use our hosted MCP server - no installation required, always up-to-date.
Endpoint: https://mcp.templated.io/mcp (OAuth login) — or https://mcp.templated.io/mcp?apiKey=YOUR_API_KEY for automation
Claude (web, desktop, mobile)
Templated is listed in the Claude connectors directory: open claude.ai/directory/templatedio (or Customize → Connectors and search for Templated), click Connect and sign in to your Templated account. To add it as a custom connector instead, use Settings → Connectors → Add custom connector with https://mcp.templated.io/mcp.
Claude Code
claude mcp add --transport http templated https://mcp.templated.io/mcpThen run /mcp inside Claude Code and sign in when prompted.
Cursor IDE
Or add it manually to ~/.cursor/mcp.json. Either way, click Login when Cursor asks and sign in to Templated:
{
"mcpServers": {
"templated": {
"url": "https://mcp.templated.io/mcp"
}
}
}For automations, use https://mcp.templated.io/mcp?apiKey=your-api-key-here as the URL instead.
ChatGPT
Go to Settings → Connected Apps → Add MCP Server
Enter URL:
https://mcp.templated.io/mcpSet Authentication to OAuth and sign in to Templated when prompted
Click Create
Alternative (automations): enter https://mcp.templated.io/mcp?apiKey=your-api-key-here and set Authentication to No Auth instead.
Plugin for Cursor and Grok
This repository is also a plugin that bundles the hosted MCP server with three skills:
templated: render from existing templates (exact layer names, batches, sizes, multi-page PDFs and merging PDFs)templated-design: design a new template from a short brief, or fix the layout of an existing onesetup: a first-run flow that checks the account and produces a first render, including on accounts with no templates yet
File | Used by |
| Cursor and Grok Bot (Cursor Marketplace) |
| Grok Build (xAI plugin marketplace) |
| The MCP server ( |
| The skills above (also packaged in the ChatGPT plugin) |
No configuration is needed: the first tool call opens the Templated sign-in (OAuth), where you can also create an account. Nothing runs on your machine; the plugin only connects to the hosted server.
To try it locally in Cursor, clone this repository into ~/.cursor/plugins/local/templated and run Developer: Reload Window.
Option 2: Local Server
Run the MCP server locally on your machine. Requires Node.js 18+.
Claude Desktop
To run the server locally instead of using the remote connector, add to your config file:
macOS: ~/Library/Application Support/Claude/claude_desktop_config.json
Windows: %APPDATA%\Claude\claude_desktop_config.json
{
"mcpServers": {
"templated": {
"command": "npx",
"args": ["mcp-server-templated"],
"env": {
"TEMPLATED_API_KEY": "your-api-key-here"
}
}
}
}Cursor IDE (Local)
Add to ~/.cursor/mcp.json:
{
"mcpServers": {
"templated": {
"command": "npx",
"args": ["mcp-server-templated"],
"env": {
"TEMPLATED_API_KEY": "your-api-key-here"
}
}
}
}Manual Installation
# Using npx (no install needed)
npx mcp-server-templated
# Or install globally
npm install -g mcp-server-templatedAuthentication
The remote server supports two ways to authenticate:
OAuth 2.1 via the Templated authorization server — used automatically by interactive clients (Claude, ChatGPT, Cursor, Claude Code) when you connect to
https://mcp.templated.io/mcpwithout an API key.API key via
?apiKey=YOUR_API_KEYon the URL or anAuthorization: Bearer YOUR_API_KEYheader — used for automations and multi-tenant scoping (folderId/externalId).
Available Tools
Full reference with inputs, permission class (read, write or sensitive write) and side effects: docs/tools.md (regenerate with npm run build && npm run docs:tools).
Render Operations
Tool | Description |
| Create an image, video, or PDF from a template |
| Get details of a specific render |
| List all renders in your account |
| Delete a render |
| Merge multiple PDF renders into one |
Template Operations
Tool | Description |
| List all templates (with search/filter) |
| Get template details |
| Get all layers of a template |
| Get pages of a multi-page template |
| Create a new template (single-page via |
| Update an existing template (add or edit pages via |
| Clone a template (optional new |
| Delete a template |
| List renders from a template |
Folder Operations
Tool | Description |
| List all folders |
| Create a new folder |
| Rename a folder |
| Delete a folder |
Asset Operations
Tool | Description |
| List uploaded assets |
| Upload a file from URL |
| Delete an upload |
| List custom fonts |
| Upload a custom font |
| Delete a custom font by name |
Account
Tool | Description |
| Get render usage for the current period |
In-chat render preview
In hosts that support MCP Apps (Claude on web, desktop and mobile, ChatGPT, VS Code and others), create_render and get_render show their result as a preview card inside the conversation:
Images (PNG, JPG, WebP) appear fitted to the card; MP4 renders get a video player; PDFs show as a file tile.
Multi-page templates show their pages as a grid (a list for PDFs).
Each card has Download and Open in editor buttons, and follows the host's light or dark theme.
A render that is still processing shows a waiting state and updates itself when the file is ready.
The card is the UI resource ui://templated/render-card-v1.html, one self-contained HTML file (about 21 KB) that loads nothing except the render files from cdn.templated.media and templated-assets.s3.amazonaws.com. Clients without MCP Apps support (Cursor, Claude Code, scripts) get the same JSON text result as before, plus structuredContent with the render's URL, format, size and status.
Example Usage
Once configured, you can ask your AI assistant to:
"List my templates"
"Create a render from template [ID] with the title set to 'Hello World'"
"Generate a PDF certificate with the name 'John Doe'"
"What layers does template [ID] have?"
"Create a new template called 'Social Post' with a title and image layer"
"Clone my Instagram template and rename it"
"Upload this image to my account"
"Show my API usage and account info"
Creating a Render
"Create a render using template abc-123 with these changes:
- Set the 'title' layer text to 'Welcome!'
- Set the 'photo' layer image to https://example.com/photo.jpg
- Output as PNG with transparent background"Claude will use the create_render tool with:
{
"template": "abc-123",
"format": "png",
"transparent": true,
"layers": {
"title": { "text": "Welcome!" },
"photo": { "image_url": "https://example.com/photo.jpg" }
}
}Development
# Clone the repository
git clone https://github.com/templated-io/mcp-server-templated.git
cd mcp-server-templated
# Install dependencies
npm install
# Build (the server, then the preview card into dist/ui/render-card.html)
npm run build
# Test
npm test
# Run in stdio mode (for local MCP clients)
TEMPLATED_API_KEY=your-key node dist/index.js
# Run in HTTP mode (for remote access)
PORT=3456 node dist/index.js
# Then access: http://localhost:3456/mcp?apiKey=your-keyTesting with MCP Inspector
npx @modelcontextprotocol/inspector npx mcp-server-templatedResources
License
MIT
Available Tools
25 toolsclone_templateCInspect
Create a copy of an existing template
Inputs: template_id (required), name.
create_folderBInspect
Create a new folder to organize templates
Inputs: name (required).
create_renderAInspect
Create a render (image, video, or PDF) from a template. This is the main tool for generating content. Supports formats: jpg, png, webp, pdf, mp4.
Inputs: template (required), fps, cmyk, name, +9 more.
create_templateAInspect
Create a new template programmatically with layers. IMPORTANT: Each layer must have a 'layer' field (unique identifier/name), not 'name'. Valid layer types are: 'text', 'image', 'shape', 'rating'. Use 'shape' for rectangles, circles, and other shapes - shapes require an 'html' field with SVG content. For a multi-page or multi-size template (several sizes in one template), pass 'pages' instead of 'layers': each page carries its own width/height and its layers as an object keyed by layer name.
Inputs: name (required), width (required), height (required), pages, +3 more.
create_uploadCInspect
Upload a file from a URL
Inputs: url (required), name.
delete_folderCDestructiveInspect
Delete a folder
Inputs: folder_id (required).
delete_fontBDestructiveInspect
Delete a custom font
Inputs: font_name (required).
delete_renderCDestructiveInspect
Delete a specific render
Inputs: render_id (required).
delete_templateCDestructiveInspect
Delete a template
Inputs: template_id (required).
delete_uploadCDestructiveInspect
Delete an uploaded asset
Inputs: upload_id (required).
get_accountARead-onlyInspect
Get the connected account's render usage for the current period (renders used and renders included)
Inputs: no named parameters.
get_renderARead-onlyInspect
Retrieve a specific render by its ID to get the status and file URL
Inputs: render_id (required).
get_templateBRead-onlyInspect
Retrieve a specific template by ID
Inputs: template_id (required).
get_template_layersARead-onlyInspect
Get all layers of a template. Use this to understand what layers can be modified when creating a render.
Inputs: template_id (required).
get_template_pagesBRead-onlyInspect
Get all pages of a multi-page template
Inputs: template_id (required).
list_foldersBRead-onlyInspect
List all folders in the account
Inputs: page, limit.
list_fontsBRead-onlyInspect
List all custom fonts uploaded to the account
Inputs: page, limit.
list_rendersCRead-onlyInspect
List all renders in the account
Inputs: page, limit.
list_template_rendersBRead-onlyInspect
List all renders created from a specific template
Inputs: template_id (required), page, limit.
list_templatesARead-onlyInspect
List all templates in the account. Use this to find template IDs for rendering.
Inputs: page, tags, limit, query, +3 more.
list_uploadsBRead-onlyInspect
List all uploaded assets (images, videos)
Inputs: page, limit.
merge_rendersBInspect
Merge multiple PDF renders into a single PDF document
Inputs: render_ids (required), host.
update_folderCDestructiveInspect
Update a folder's name
Inputs: name (required), folder_id (required).
update_templateADestructiveInspect
Update an existing template. IMPORTANT: Each layer must have a 'layer' field (unique identifier), not 'name'. Valid types: 'text', 'image', 'shape', 'rating'. For multi-page templates use 'pages': a page name that does not exist yet is ADDED to the template, an existing one has its layers merged (or replaced with replaceLayers). To change page sizes on a multi-size template set width/height per page inside 'pages'; top-level width/height are rejected there because they would resize every page.
Inputs: template_id (required), name, pages, width, +6 more.
upload_fontBInspect
Upload a custom font from a URL
Inputs: url (required), name (required).
Tool Schema Changelog
Recent tool additions, removals, and schema changes observed during successful MCP inspections.
Loading tool schema changelog…
TDQS
Scored across 25 tools
Each tool targets a distinct resource+action (renders, templates, folders, uploads, fonts, account). No two tools appear to do the same thing; create_render vs merge_renders vs list_template_renders are clearly separated. Minor potential overlap between list_renders and list_template_renders, but descriptions distinguish them.
Mostly consistent verb_noun snake_case (create_render, list_templates, delete_folder). Minor deviations: create_upload vs upload_font use different verbs for upload actions, and pluralization varies (get_render vs list_renders). Still highly readable and predictable.
25 tools is heavy for a single MCP server; the surface spans six sub-domains. While each tool is distinct, the set could likely be consolidated (e.g., folder CRUD, font CRUD). At the 25-tool threshold, it is borderline rather than well-scoped.
Core lifecycle for templates (create/get/update/delete/clone) and renders (create/get/list/delete/merge) is covered. Gaps exist for uploads and fonts (no get/update, only list/create/delete) and folders lack a get tool, but agents can work around via list calls.
Maintenance
Related MCP Connectors
Create images, video and audio; reuse assets and build Canvas workflows from your AI assistant.
Render on-brand images, PDFs and videos from templates; create and edit templates and brands.
1Create AI images and videos, manage projects and credits, and use workspace campaign context.
Image and video AI tools and your own pipelines, run from any AI assistant.
Related MCP Servers
- AlicenseNot gradedqualityDmaintenanceEnables programmatic design creation and manipulation on the Canvelete platform through AI assistants. Supports design management, canvas element manipulation, template application, asset management, and real-time synchronization with the design editor.26 npmMIT
- AlicenseAqualityCmaintenanceEnables AI assistants to generate AI avatar videos, manage templates, and work with assets through natural language commands via the HeyGen API.7MIT
- AlicenseNot gradedqualityBmaintenanceEnables AI assistants to manage Cloudinary media assets, including uploading, searching, transforming, and organizing images and videos through natural language.950 npmMIT
- AlicenseNot gradedqualityCmaintenanceEnables AI assistants to generate PDFs from templates using dynamic data, manage async PDF jobs, list templates, and check account credits and transaction history through natural language.66 npmMIT