Point your agent at one URL and start building popups.
OptiMonk’s popup builder is operable by an AI agent. Give it a reference image or a description and it builds a working popup for a specific store, edits it in conversation, and turns it into a campaign draft.
Supported OAuth clients connect using the URL below, without an API key. See your client guide for prerequisites and authentication.
MCP endpoint
https://agentic.optimonk.com/mcp
Which client are you using?
Pick your client for configuration, authentication, and a connection check:
Claude — web, desktop, or Claude Code. See Claude.
Cursor — see Cursor.
ChatGPT or Codex — see ChatGPT.
Other MCP clients — check OAuth compatibility or use an API key. See Other MCP clients.
Or just tell your agent
If your agent can set up its own tools, paste this and let it do the work:
Prompt
Help me connect to the OptiMonk popup builder MCP server at https://agentic.optimonk.com/mcp. Use the client setup guide linked from the OptiMonk Quickstart and check whether my client supports its OAuth flow. Guide me through authentication, or the API-key fallback if needed. Once connected, call whoami and tell me which OptiMonk account you are acting as and what you are allowed to do.
Then try it
“Build me an exit-intent discount popup for example-shop.com”
“Here's a screenshot of a popup I like — recreate it for my store”
“Make the headline shorter and the button green”
More in What you can ask it. Builds are asynchronous, so a good agent will wait rather than poll — the watch_job tool exists for exactly that.
Scripting instead of an agent?
There is a REST API behind the same credentials, described by an OpenAPI document at /.well-known/openapi.json. Create an API key under Settings → API & Connected apps and see Authentication & access.