1. Describe it
Tell us what you want in plain English. No prompt-engineering rituals required.
Agents shouldn't
need stock photos.
Drop Get Images into Claude, Cursor, or any MCP-aware agent. Your agent gains a tool to generate exactly the image it needs — no SDK, no glue code.
Works with
- Claude
- Cursor
- VS Code
- OpenAI
- Gemini
claude_desktop_config.json
{
"mcpServers": {
"get-images": {
"type": "http",
"url": "https://mcp.get-images.dev/mcp",
"headers": {
"Authorization": "Bearer gi_live_••••••••"
}
}
}
}Paste it, restart your agent, and you're wired in.
Live demo
Watch an agent use it.
Your coding agent calls Get Images like any other MCP tool. Here's what that looks like end to end — terminal on the left, the page being built on the right.
~/projects/my-app — claude-code
❯ claude
Type your message…
my-app.com
MyApp
ProductPricingSign up
Build something great.
The fastest way to ship your idea this weekend.
hero image
How it works
Three steps. That's it.
Built for vibe coders and agentic builders. No SDK to learn, no dashboard hopping — just images where your agent already lives.
2. Pick a model
Choose from OpenAI, Gemini, and more. Switch any time — your prompts stay portable.
3. Plug into your agent
One MCP endpoint. Drop the key into Claude, Cursor, or anything that speaks MCP.