@spronta/mcp
MCP server for uploading, transforming, and serving images on a global CDN — directly from AI assistants.
Connect Claude, GPT, Cursor, Windsurf, or any MCP-compatible client to the Spronta Image CDN. Upload images from URLs or base64, apply real-time transforms (resize, crop, format conversion, smart crop, blurhash), create reusable presets, generate signed URLs, and monitor usage — all through natural language.
Why use this?
- Upload images from AI — give your LLM an image URL and it uploads, optimizes, and returns a CDN link
- Real-time transforms — resize, crop, convert to WebP/AVIF/JXL, smart crop with face detection
- Global CDN delivery — images served from edge locations worldwide
- Blurhash generation — automatic blur placeholders computed on upload
- Signed URLs — HMAC-SHA256 signed URLs with expiration for private images
- Named presets — create reusable transform configurations (e.g.
thumbnail, hero, og-image)
- No vendor lock-in — MIT licensed, open source
Quick start
Install
npm install -g @spronta/mcp
npx @spronta/mcp
Claude Code
claude mcp add spronta \
-e SPRONTA_API_KEY=spronta_img_... \
-e SPRONTA_PROJECT_ID=your-project-id \
-- npx @spronta/mcp
Claude Desktop / Cursor / Windsurf
Add to your MCP configuration file:
{
"mcpServers": {
"spronta": {
"command": "npx",
"args": ["@spronta/mcp"],
"env": {
"SPRONTA_API_KEY": "spronta_img_...",
"SPRONTA_PROJECT_ID": "your-project-id"
}
}
}
}
Get your API key
- Sign up at app.spronta.com (free tier: 100 images)
- Create a project in the dashboard
- Copy your API key from project settings
Environment variables
| Variable | Required | Description |
|---|
SPRONTA_API_KEY | Yes | Your project API key |
SPRONTA_PROJECT_ID | No | Default project ID (can also be passed per-tool) |
SPRONTA_API_URL | No | API base URL (default: https://app.spronta.com/api) |
Projects
| Tool | Description |
|---|
list_projects | List all image projects |
create_project | Create a new project |
get_project | Get project details with usage stats |
update_project | Update name or custom domain |
delete_project | Permanently delete a project |
Upload & Images
| Tool | Description |
|---|
upload_image | Upload from URL or base64 — handles presigned upload, blurhash generation, and CDN URL creation in one call |
list_images | List images with pagination |
update_image | Update alt text and tags |
delete_image | Delete an image from CDN and storage |
| Tool | Description |
|---|
list_presets | List named transform presets |
create_preset | Create a preset (use in CDN URLs with ?t=name) |
update_preset | Update a preset's name or transforms |
delete_preset | Delete a preset |
URL Signing
| Tool | Description |
|---|
get_signing_config | Get URL signing configuration |
update_signing | Enable/disable HMAC-SHA256 signing |
generate_signed_url | Generate a signed CDN URL with optional expiration |
Analytics & Utility
| Tool | Description |
|---|
get_usage | Get daily metrics (requests, bandwidth, transforms) |
build_cdn_url | Build a CDN URL with transform parameters (no API call) |
Example prompts
Once connected, just ask your AI:
> Upload this image to my project: https://example.com/photo.jpg
> Create a thumbnail preset: 200x200, cover crop, face detection, webp format
> How many requests did my project handle this week?
> Generate a signed URL for hero.jpg that expires in 1 hour
> List all my images and update the alt text on the product shots
> Build a CDN URL for banner.png at 1200px wide in avif format
When creating presets or building URLs, these transforms are available:
| Parameter | Type | Description |
|---|
width | integer | Output width (1–8192) |
height | integer | Output height (1–8192) |
fit | string | cover, contain, fill, scale-down, crop, pad |
format | string | auto, webp, avif, jpeg, png, jxl |
quality | integer | 1–100 |
qualityMode | string | high, medium, low (perceptual targeting) |
gravity | string | auto, face, center, position keywords |
blur | integer | Gaussian blur (1–250) |
sharpen | number | Unsharp mask (0–10) |
radius | integer | Corner radius (1–9999, or "max" for circle) |
grayscale | boolean | Convert to grayscale |
brightness | integer | -100 to 100 |
contrast | number | -100 to 100 |
saturation | integer | -100 to 100 |
sepia | boolean | Sepia tone filter |
rotate | integer | 90, 180, 270 |
flip | string | h (horizontal), v (vertical) |
How upload works
The upload_image tool handles the full upload pipeline in a single call:
- You provide an image URL (or base64 data) and a filename
- The server fetches the image
- Requests a presigned upload URL from Spronta
- Uploads the file directly to storage
- Confirms the upload — Spronta computes the blurhash and detects dimensions
- Returns the confirmed image with CDN URL, blurhash, and metadata
Links
License
MIT