Whatagraph MCP (Model Context Protocol) lets your AI assistant — Claude, ChatGPT, and others, directly read your Whatagraph account data: report sources, metrics, account structure, and more.
Once connected, you can ask natural-language questions about your data, spot broken sources, and even generate visualizations, all without leaving your AI chat.
Before you start
You will need the right access level before setting this up:
On Claude, you need Owner or Full Admin access to your Claude workspace.
On ChatGPT, you need Developer access to add custom apps.
If you are on a team plan and do not see the option to add a custom connector, reach out to whoever manages your account and ask them to complete this setup.
Also, make sure you are logged into your Whatagraph account in the same browser before you begin, it will make the authorization step seamless.
How to connect in 5 steps
Step 1. Open Settings in Claude or ChatGPT and navigate to Connectors. Look for the option to add a custom connector.
Step 2. Paste the Whatagraph MCP server URL into the custom connector field: https://mcp.whatagraph.com/mcp
Step 3. Save and authorize. Your AI platform will prompt you to connect your Whatagraph account. As long as you're already logged into Whatagraph in the same browser, this is a one-click step.
Step 4. After the connection is authorized you will be able to see the list of permissions that the connector has.
Step 5. Test the connection by asking: "What do you have access to in my Whatagraph account?" — your assistant should respond with your connected sources and account details.
Note: If permissions are set to "Needs approval", you'll need to authorize each time the connector pulls data. To skip that, switch to "Always allow" in the permissions page.
Step 6. (Optional) Download skills from GitHub and upload them under Customize
Does this work on ChatGPT too?
Yes. The process is nearly identical to Claude. Use the same MCP server URL and follow the same connection steps. The only difference is that you need Developer access on your ChatGPT account to set up custom apps.
What you can do once connected
Once connected, your AI assistant can answer questions about your Whatagraph data in seconds. A few things to try straight away:
Daily performance briefing — ask for a live summary across your top clients without opening a single dashboard
Broken source detection — surface stale or disconnected sources before they become a client problem
Goal pacing check — see which active goals are on track and which need attention, mid-month
Spend analysis — get a full cross-channel spend breakdown for any time period, no report building required
Account health snapshot — check active sources, report issues, and upcoming automations all in one ask
Tip: These prompts work best when your sources are organized into Spaces by client.
Go further with ready-made skills
Think of them as pre-built templates for the most common agency workflows, spend analysis, client briefings, or source audits. So you're not writing prompts from scratch each time.
Pre-built prompt templates that act like a customer success manager, helping you get more out of your data every day.
You can find and download them from GitHub, then upload them to your AI platform under Customize → Skills.
Here are some things you can do with skills loaded:
Ask how much you have spent across channels in the last 12 months
Get aggregated performance summaries across all connected sources
Identify broken or disconnected data sources
Request charts and visual summaries of your reporting data
To use a skill, type / in your chat and reference the skill by name. For example, with the spend analysis skill loaded, just ask: "How much have I spent in the last 12 months?"
Prefer to see it in action? Watch the full video walkthrough below.