Introduction

This article covers using Webdock’s remote MCP server as a custom MCP app/connector in ChatGPT.

Webdock MCP URL: https://mcp.webdock.io

Webdock authenticates with OAuth. When ChatGPT connects, you sign in to Webdock and approve the permissions the app may use.

Availability

OpenAI’s MCP/app capabilities and plan availability are evolving quickly.

At the time this draft was written, OpenAI documents full custom MCP support, including write/modify actions, for ChatGPT Business, Enterprise and Edu workspaces. OpenAI also documents more limited MCP use on some other plans.

Because Webdock MCP is intended to perform real infrastructure actions as well as reads, check OpenAI’s current plan requirements before publishing this article.

Add Webdock as a custom app

Follow these steps in the ChatGPT app.

1. Open the app settings
2. In the sidebar, head over to Plugins
3. Then click Add -> “Add MCP Server”
4. Enter name, such as: Webdock
5. Set type to Streamable HTTP
6. For the URL enter the Webdock remote MCP URL: https://mcp.webdock.io and save it (do not worry about the other fields)
7. Complete the Webdock OAuth flow by clicking on the Authorize button

Exact menu labels differ between Business and Enterprise/Edu and may change over time.

Use Webdock in a chat

Open up a normal conversion and start with a read-only test:

Use Webdock to show me the server locations and profiles available to my account. Do not create anything.

Then:

Show me my current Webdock servers and summarize their state without making changes.

Prompt for native Webdock workflows

If you want ChatGPT to use a Webdock-managed capability rather than building an equivalent solution manually, say so.

Example:

Create an inexpensive WordPress server in Denmark using Webdock’s managed WordPress tooling. Inspect profile capabilities first, choose the cheapest compatible profile, show me the monthly price, and ask before creating the billable server.

LAMP and LEMP

For PHP applications, explicitly tell ChatGPT when you want a ready Webdock web stack:

Use Webdock’s LEMP image because this application needs Nginx, PHP and database components. Do not start from a bare OS unless there is a good reason.

This helps the agent avoid spending time reconstructing a stack that Webdock already provides.

Snapshot before large changes

For production or important servers, consider prompting:

Create a Webdock snapshot before any major package, web-server, PHP or database changes. Wait for the snapshot to complete before continuing.

A snapshot is a useful recovery point, not a replacement for your normal backup strategy.

Use a strong model

Infrastructure automation requires more planning and tool discipline than ordinary Q&A.

Use a capable model for provisioning, troubleshooting and multi-step changes. Lightweight models are more likely to select the wrong tool, lose track of asynchronous operations or improvise unsafe workarounds.

For read-only inventory questions, smaller models may be sufficient.

Example prompts

Create a managed WordPress server

Create a new inexpensive WordPress server in Denmark using Webdock’s managed WordPress functionality. Filter profiles by WordPress capability first and then choose the cheapest eligible profile. Show the displayed monthly price before provisioning. Use a suitable LEMP image and set the site title to “Hello World”.

Plan without changing anything

I want to deploy a PHP API on Webdock. Inspect available Denmark profiles and images and recommend the cheapest sensible LEMP configuration. Do not provision anything.

Snapshot and upgrade

On my-server, create a snapshot and wait for it to finish. Then explain the PHP upgrade you plan to perform and ask me before making the change.

Current OpenAI-specific notes

At the time of writing:

• full custom MCP write/modify support is documented for Business, Enterprise and Edu;
• workspace admins/owners control custom app availability;
• important write actions may trigger ChatGPT confirmation;
• app tool updates may need to be refreshed and reviewed in ChatGPT;
• OpenAI’s Agent mode and Deep Research have separate app/tool limitations.

These product details change frequently. Verify them against OpenAI’s current documentation before publication.

Please reach out to Webdock Support if you need assistance.