This guide walks you through connecting an AI agent to Get Pronto using the Model Context Protocol (MCP). Once set up, your agent can upload images, generate transformed URLs, and manage files — all through natural language.

The MCP server works with any MCP-compatible client, including Claude Code, Claude Desktop, Cursor, and Windsurf.

Prerequisites

An MCP-compatible AI client (Claude Code, Cursor, Windsurf, etc.)
Node.js 18+
Optionally, a Get Pronto account and API key (not required — agents can generate ephemeral test keys)

1. Add the MCP server

Claude Code

Run this in your terminal:

claude mcp add getpronto -- npx -y getpronto-mcp

To use an existing API key:

claude mcp add getpronto -e GETPRONTO_API_KEY=pronto_sk_... -- npx -y getpronto-mcp

Claude Desktop, Cursor, and Windsurf

Add the following to your MCP configuration file:

{
  "mcpServers": {
    "getpronto": {
      "command": "npx",
      "args": ["-y", "getpronto-mcp"]
    }
  }
}

To use an existing API key, add an env block:

{
  "mcpServers": {
    "getpronto": {
      "command": "npx",
      "args": ["-y", "getpronto-mcp"],
      "env": {
        "GETPRONTO_API_KEY": "pronto_sk_..."
      }
    }
  }
}

Where is the config file? Claude Desktop: ~/Library/Application Support/Claude/claude_desktop_config.json (macOS) or %APPDATA%\Claude\claude_desktop_config.json (Windows). Cursor and Windsurf: check their MCP documentation for the config file location.

2. Verify the connection

Ask your agent something like:

List my Get Pronto files

If no API key is configured, the agent will automatically generate an ephemeral test key on the first request. You'll see it call the generate_test_key tool before proceeding.

3. Upload an image

Your agent can upload images from file paths, URLs, or data URLs:

Upload ./assets/hero.png to Get Pronto

Upload this image from https://example.com/photo.jpg

The agent uses the upload_image tool, which handles the presigned URL flow behind the scenes. It returns the file ID, URL, and metadata.

4. Transform images

Ask the agent to generate transformed image URLs:

Resize my latest upload to 1200x630 and convert to WebP

Create a blurred thumbnail of file abc123 at 200x200

The transform_image tool supports:

Resize — width, height, and fit mode (cover, contain, fill, inside, outside)
Format — jpeg, png, webp, avif
Quality — 1 to 100
Blur — Gaussian blur (0.3 to 1000)
Sharpen and grayscale
Rotate — -360 to 360 degrees
Crop — x, y, width, height
Border — width and hex color

The returned URL is a CDN-cached transform URL. The first request generates the image; subsequent requests are served from the edge.

5. Manage files

Your agent can list, inspect, and delete files:

List all my uploaded files

Get the details for file abc123

Delete all files in the "temp" folder

The list_files tool supports pagination and folder filtering. The get_file tool returns the full file metadata including URL, dimensions, file size, and MIME type.

Ephemeral test keys

If you don't configure an API key, the agent calls generate_test_key to create a temporary account instantly. This is ideal for trying things out without signing up.

Ephemeral keys have the following limits:

Limit	Value
Storage	100 MB
Monthly bandwidth	500 MB
Monthly transforms	100
Max file size	10 MB (images only)
TTL	7 days

After 7 days, the ephemeral account and all its files are automatically deleted. To keep your files and unlock higher limits, create a free account and use your secret API key instead.

Real-world examples

Here are some practical ways to use Get Pronto with AI agents:

Generate Open Graph images for a project:

Upload hero.png, resize to 1200x630 as WebP, and give me the URL for my meta tags

Batch process screenshots:

Upload all PNGs in ./screenshots, convert each to WebP at 85% quality, and list the URLs

Quick image hosting while coding:

I need to host this image for my README — upload it and give me a markdown image tag

Prototype an image gallery:

List my files and generate 200x200 cover-fit thumbnails for each one

Environment variables

Variable	Description	Default
`GETPRONTO_API_KEY`	Your Get Pronto secret API key	None (agent uses `generate_test_key`)
`GETPRONTO_BASE_URL`	API base URL	`https://api.getpronto.io/v1`

Next steps

MCP documentation — full reference for all tools and configuration
SDK documentation — use the SDK directly in your own applications
Image transforms reference — all available transformation parameters
Add Image Hosting to Your Next.js App — integrate Get Pronto into a web application

Connect AI Agents to Get Pronto with MCP