The future of data collection starts here
SurveyCTO is delighted to be one of the first in the data collection platform space to introduce a hands-on AI copilot: the SurveyCTO AI agent skill!
The SurveyCTO AI skill and MCP server are the latest creations in our family of principle-driven AI innovations. They are intended to make SurveyCTO tasks—from form programming to creating field plug-ins to translating forms into different languages and much more—highly accessible to users at all levels of technical skill through the power of AI. It’s available now in public beta, ready to be configured and used to transform how easily you can create surveys and workflows with SurveyCTO.
For best results, we recommend using the skill with our MCP server.
Here’s a quick breakdown of what AI agent skills and MCP servers are, and how they work:
- An AI agent skill supplies generalized AI agents with deep, specialized knowledge about an external platform. In the case of SurveyCTO, this means our agent skill will give AI agents SurveyCTO domain expertise in designing, editing, and debugging forms, server dataset XML definitions, Data Explorer workbook definitions, and field plug-ins. Instead of giving generic advice from the internet, the agent becomes a SurveyCTO expert that gives tailored, highly-relevant responses.
- An MCP (Model Context Protocol) server provides a suite of tools for AI agents to access so that when you’re conversing with them, they don’t simply tell you what to do—they actually do what you want to do.
Think of the skill as the expert consultant, and the MCP server as the consultant’s hands. Without the MCP server, our AI agent can only give you instructions on how to build a SurveyCTO form; it might also produce a generic form for you, but that form may have inconsistencies that result in invalid form design. The magic happens when you use the skill and MCP server together, resulting in the output you want, without you having to necessarily make extra adjustments so that they can work in the SurveyCTO platform later on.
Table of Contents
What our AI agent skill and MCP server can do
In combination with our new MCP server, the agent skill enables users to use natural-language prompts and engage in secure “vibe coding” with popular AI agents powered by models like Claude and GPT, and have them complete SurveyCTO tasks that normally take hours or days.
Yes, that’s right: Language-prompted, AI-generated data collection instruments are here—and SurveyCTO is bringing it to you!
The skill paired with the MCP server can perform many SurveyCTO tasks. Here are some actions you can ask it to do:
- Create SurveyCTO forms based on natural language descriptions
- Easily design field plug-ins that customize how survey fields look and function
- Translate your form into other languages
- Convert non-SurveyCTO forms into SurveyCTO forms
We’re also continuing to refine the skill & MCP server to make the following possible:
- Generate dataset definitions
- Link datasets across multiple forms to create connected, end-to-end workflows and custom applications
- Build data monitoring dashboards in the Data Explorer populated with exactly what you want to visualize
- And much, much more!
Now, here’s how to set this all up for yourself.
How to get the skill and MCP server set up
Follow these steps to configure both the agent skill and the MCP server.
Want a video walking through how to set this up? Watch our demo webinar.
- Download the SurveyCTO agent skill on GitHub—and read carefully through all information, especially the Readme, since it contains lots of important details on how the skill works
- Upload the package to your AI agent of choice! Each agent will work a little differently. In Claude Cowork, uploading a skill looks like this:
3. Next, add the MCP server (strongly recommended for best results) using this address: https://assistant-be.surveycto.net/mcp. Add the server to your AI agent as a custom connector once the skill is uploaded, and name it “SurveyCTO tools”:
For ease of use, we recommend you set Tool Permissions to Always allow, so that the agent doesn’t ask you for permission to execute every single action.
- Once this is all connected, you can start chatting with your AI agent and ask it to perform a SurveyCTO task for you, like below:
Important information before you get started
The advent of artificial intelligence led to an exciting time in survey research and data-gathering work. We are looking forward to the ways that our AI agent skill makes SurveyCTO a better tool for a wider variety of researchers, companies, and organizations.
While this development is exciting, we also want to emphasize a few points:
- The SurveyCTO AI agent skill is a work in progress—it can make mistakes! Always include a layer of human review for outputs before using them in your work, as you should with any AI tool.
- Because agent output must eventually be deployed within the SurveyCTO platform with its data security features, using our agent skill is still much more secure than using standard “vibe coding” platforms.
- We want to hear from you! If you built something you’re excited about, tag us on LinkedIn. Similarly, if you have feedback for us, email and let us know—we look forward to hearing your thoughts and seeing ways to improve AI innovations at SurveyCTO.
- Since the skill is in public beta, updates will periodically occur. Whenever an update happens, you will need to revisit our Github repository, download the updated skill, and go through the configuration process again. The easiest way to get updates is to sign up for a free GitHub account and select the Watch setting to get notifications.
To update the skill in an AI agent, simply replace it (see screenshot below)! No need to delete the current installed skill.
While configuration and usage of our AI agent skill is straightforward, there’s also plenty of greater detail to dive into in order to get the most out of it.
Check out the following to get started on your deep dive:
➡️ Watch our live demo of the AI agent skill
➡️ Learn about how we approach AI at SurveyCTO
➡️ Access our MCP server here
➡️ Get the GitHub skill package here
