Introduction
MCP (Model Context Protocol) is a standard that lets AI tools connect to external data sources and use them directly in a conversation. The Attest MCP server uses this to let AI tools, such as Claude Desktop, query your Attest survey results without you writing any code or setting up a data connection.
It's aimed at developers and AI practitioners who want to bring Attest results into an AI-native workflow, such as an agent, assistant, or internal tool.
Note: The MCP server is currently available to a smaller cohort of customers ahead of a wider release. Speak to Sales or the Customer Team for more information.
How it works
Connecting the Attest MCP server
In your Attest dashboard, go to your Profile & Setting page. (This is the same type of credential used for the Attest Results API, so if you already have one, you can reuse it)
Generate an API key and copy it somewhere safe
Add the Attest MCP server using this Setup guide based on your AI tool, using your API key to authenticate.
Once connected, the Attest MCP tools appear in your AI tool, ready to use.
Each user has a single API key at a time. If you generate a new key, for example because your existing key has been compromised, this immediately invalidates the old one. There's currently no way to view or explicitly revoke a key separately, so generating a new one is how you rotate access.
Full setup steps are in the Attest developer documentation.
AI tool | How you connect |
Claude Desktop (Mac/Windows app) | Install the Attest extension file and paste in your API key |
ChatGPT | Add Attest as a connector in Settings → Connectors (needs Developer mode) |
Microsoft Copilot Studio | Add Attest as an MCP tool to a Copilot Studio agent |
GitHub Copilot (VS Code or CLI) | Add Attest to your MCP config file |
Claude Code | Add Attest via a terminal command |
Gemini CLI | Add Attest to your Gemini settings file |
Other MCP tools | Add |
Querying your results
Once connected, you or your AI agent can ask for Attest data in plain language, for example "show me the raw responses for our Q1 brand awareness tracker." The MCP server fetches the underlying data and returns it as structured data your AI tool can work with.
Common use cases
Conversational access to results. Query survey data in natural language, without needing to know the underlying API schema or write integration code.
Synthesis across data sources. Bring Attest results together with other research inputs, such as user interviews or usability testing, within a single AI context, to identify patterns across sources.
Tracking changes over time. Use an AI agent to help reconcile or compare question phrasing and results across tracker waves.
Powering internal tools. Build assistants or internal query tools that let colleagues ask questions about Attest results without logging into the platform.
Cross-survey analysis. Group a subset of your surveys by category, theme, or any criteria that matters to you, and analyse them together to spot trends a single survey wouldn't show.
Scope of release
Access is read-only. The MCP server cannot create or edit boards, segments, filters, or key findings, cannot draft or amend surveys, and cannot change any admin settings.
Attest Measure only. Explore and AI-moderated interviews, including transcripts, are not currently supported.
Published results only. Draft studies are not visible.
Currently, the MCP server only supports fetching raw response data for a study.
FAQs
What is MCP?
MCP (Model Context Protocol) is a way of connecting AI tools to external data sources, so an AI assistant can query and use that data directly in a conversation.
Which AI tools can I use this with?
Claude Desktop, ChatGPT, Microsoft Copilot Studio, GitHub Copilot, Claude Code and Gemini CLI. Other tools work too if they let you add a remote MCP server with an API key.
The MCP server currently only works via mcp-remote and an API key integration. We do not yet support OAuth login for our MCP server and so it won’t support AI Clients that are dependent only on this method of authentication.
Do I need a separate credential to use MCP?
No. It uses the same API key as the Attest Results API.
Can the MCP server change anything in my Attest account?
No. Access is read-only, it can only retrieve data.
Is this available to all Attest customers?
Not yet. It's currently available to a smaller cohort of customers ahead of a wider release. If you're interested in accessing this feature, contact the Customer Team for more information.
What can I ask the MCP server?
You can ask about anything in your published Attest results. That includes finding studies, reading results and questionnaires, filtering by demographics or wave, and comparing waves. To see the current list, ask your connected AI tool: “What can the Attest MCP server do?”