Documentation

Deploy a site
Browse documentation

MCP server

Build, publish and edit websites straight from an AI assistant. Beyond pages, your AI can create a working contact form, editable content, a booking calendar and a chatbot, then wire them to the site it just deployed.

Two ways to connect

Works withclaude.ai and ChatGPTClaude Code, Claude Desktop, Cursor and Windsurf
InstallNothing — paste one addressThe @dplooy/mcp-server npm package
Sign-inOAuth — approve in Dplooy, no key to copyAn API key from Settings → API Keys
ExtraCan deploy a folder from your computer (deploy_folder)

Both give your AI the same tools, and both work on every plan.

Connect claude.ai or ChatGPT

  1. 1In claude.ai, open Settings → Connectors → Add custom connector. In ChatGPT, add it as a connector in developer mode.
  2. 2Paste the address https://api.dplooy.com/mcp.
  3. 3You are sent to Dplooy to sign in and approve access. There is no API key to copy.
  4. 4Ask for a site — for example “Build a coffee shop landing page with a contact form and publish it”.

Custom connectors on claude.ai need a paid Claude plan, and Team or Enterprise workspaces may need an admin to enable them first.

Connect Claude Code, Cursor or Windsurf

Create a key under Settings → API Keys in your dashboard — it is shown only once. Then add the server:

Claude Code
claude mcp add -t stdio \  -e DPLOOY_API_KEY=dpk_live_... \  dplooy -- npx -y @dplooy/mcp-server
Claude Desktop, Cursor or Windsurf — MCP config
{  "mcpServers": {    "dplooy": {      "command": "npx",      "args": ["-y", "@dplooy/mcp-server"],      "env": { "DPLOOY_API_KEY": "dpk_live_..." }    }  }}

What your AI can do

Hosting & editing

deploy_websiteDeploy HTML/CSS/JS to a live URL
deploy_filesDeploy a multi-file site from inline content (remote)
deploy_folderDeploy a folder from your disk (local only)
get_project_filesRead a live site's files before editing
update_project_filesEdit only the files that changed
list_projectsList all your deployed projects
get_projectOne project with everything attached to it
delete_projectDelete a project permanently
get_accountPlan, usage, limits, and available features
get_guideFull docs for a feature — the AI reads this itself

Forms (Plus & Pro)

create_formCreate a form endpoint and get its embed snippet
get_formRead one form: settings, assignment, submission count
list_formsList your form endpoints and their usage

Data (All plans)

create_data_collectionCreate a collection from a preset
update_data_rowsEdit rows — the live site updates in about a minute
list_data_collectionsList your collections
get_data_collectionOne collection's fields, rows and public URL

Bookings (Plus & Pro)

create_booking_pageCreate a booking page and get the widget embed
get_booking_pageRead a booking page: hours, slots, capacity, services
list_booking_pagesList booking pages and pending requests

Media, chatbot & wiring

upload_mediaUpload an image and get a stable public URL
list_mediaList images in your media library
assign_to_projectConnect forms, collections and bookings to a project, or move them
unassign_from_projectDetach them from a project — nothing is deleted
get_chatbotRead your chatbot settings and knowledge before editing
configure_chatbotEnable an AI chatbot with your knowledge base (Plus/Pro)
delete_form / delete_data_collection / delete_booking_page / delete_mediaClean up empty, unattached leftovers — anything with content stays yours to delete

Security

The remote server speaks stateless Streamable HTTP, so there is no session to keep alive. Authorization is OAuth 2.1 with PKCE, and access is scoped to your account only. Review or revoke any connected app under Settings → Connections; revoke a local key under Settings → API Keys.

Your plan's limits apply exactly as they do in the dashboard — the same site count, file sizes and features.

Good to know

  • AI apps cache the tool list. If a new tool doesn't show up, remove the connector and add it again.
  • Every page your AI publishes is scanned before it goes live, like any other deploy.
  • Each connection can make 60 requests a minute. Over the remote connector, one deploy_files call takes up to 200 files and one upload_media call up to 16 images of 5 MB each.
  • More detail for every client, with limits and what it can't do: MCP server guide.