Flag any card in any report and we will look into it. For everything else, email us: a person reads it.
Email [email protected] and the reply comes back signed “MCP Analytics”. Our assistant is an AI that reads what you send and writes back. It answers from the same published material this website is built on. See what to send and what comes back.
An answer can be clear, well written and still wrong for your situation. That is the case [email protected] exists for. If anything in a reply looks wrong, or is not what you expected, forward the email there and a real person reads it and replies.
Forward the original rather than starting a new message. The thread carries your question and the reply we sent, which is what a person needs in order to see what happened.
Our assistant only writes to people who emailed one of our addresses first. If you would rather not hear from us again, email [email protected] and a person will action it. You can ask us at any time, and you do not need a reason.
What we do with the data and the files you send is set out in our Privacy Policy. The terms that govern the service are in our Terms. If you want a copy of what you have sent us, or want it deleted, email the same address.
Add the MCP Analytics server to your Claude Desktop configuration:
1. Open Claude Desktop Settings → Developer → Edit Config
2. Add the MCP Analytics server config, see our installation guide for copy-paste configs
3. Restart Claude Desktop
4. Type /mcp to verify the connection
The catalog covers these families of methods (the live list is in the docs):
• Multiple regression techniques (Linear, Ridge, Lasso, and more)
• Advanced time series forecasting (ARIMA, Prophet and others)
• Exploratory analysis (Correlation, PCA, clustering)
• Hypothesis testing (T-Test, ANOVA, Chi-Square, and others)
• Causal inference and Bayesian methods
• Customer analytics and segmentation tools
All tools include interactive visualizations and AI-narrated insights.
You have three options for providing data:
1. Public URLs: Use CSV files from GitHub directly
2. Test Datasets: Use our pre-loaded datasets with provided UUIDs
3. Upload Your Own: Ask Claude to generate an upload token, then use cURL to upload your file with encryption
Yes. What we can state about how data is handled:
• Encrypted in transit
• Data automatically retained per your plan’s window (see pricing for the retention table)
• Secure Auth0 authentication
Full details are in the Privacy Policy.
Every new account gets:
• Access to the full analysis catalog
• Interactive reports with visualizations
• AI-narrated insights
• Test datasets for experimentation
If you're experiencing connection issues:
1. Verify your API key is correct in your MCP config
2. Clear the npm cache: npx clear-npx-cache
3. Restart your MCP client (Claude Desktop, Cursor, etc.)
For the experimental OAuth2 option, clear auth cache: rm -rf ~/.mcp-auth
Yes. Each report has a shareable link that:
• Opens without signing in
• Includes interactive charts
• Shows the full analysis results
• Access window depends on your plan
MCP Analytics supports multiple data formats:
• CSV (comma-separated values)
• TSV (tab-separated values)
• Excel (.xlsx, .xls)
• JSON (JavaScript Object Notation)
• Parquet (columnar storage)
• Feather (binary format)
Where to start
One email with your data and your question. We take it from there.
MCP Analytics
Data science by email
Send us your data and your question, and we will reply.
Try it on your own data
30 days free.
Answers, analytics and custom analytics. No card, no call, nothing to install.
30 days free. →Tell us what went wrong, in your own words. We capture the page you're on automatically, so no need to describe where you are.