ds@mcpanalytics.aiEmail your data and your question and get an answer about your own data, with the method shown.Email your data and question. Get an answer.
The engine

Your agent already knows how to use us.

MCP Analytics is an MCP server (MCP, the Model Context Protocol, is the standard way AI assistants connect to outside tools). Connect it and your AI can see the tools itself: upload data, ask a question, follow a build, search past reports. No SDK to write against.

Install → The docs →

What the agent seesthe MCP toolset
▸ datasets_upload  bring the file
▸ path_resolver  picks how the question is answered
▸ path_tool_run  run a library tool on the data
▸ path_rerun  re-run on fresh data
▸ ask_library  one answer across every report
discovered automatically over MCPagent-native
C
Claude
G
ChatGPT
Cursor
Your Agent

One Config. Direct Access.

Add this to your MCP client configuration and restart.
That's the entire integration. 

{
  "mcpServers": {
    "mcpanalytics": {
      "command": "npx",
      "args": ["-y", "@mcp-analytics/mcp-analytics"],
      "env": {
        "MCP_ANALYTICS_API_KEY": "mcp_your_key_here"
      }
    }
  }
}
C
Claude Desktop
Settings > Developer > Edit Config
Cursor
Settings > MCP > Add Server
Other Clients
MCP configuration file

What Your Agent Gets

Direct access to analytics tools through standard MCP calls. The snippets below are illustrative sketches of a call, not full signatures; the docs have the exact arguments.

discover_tools

Semantic tool discovery

Find the right analysis tool using natural language. "I need to predict customer churn" returns ranked, compatible tools. 

// Example call
discover_tools({
  query: "regression analysis",
  dataset_uuid: "ds_abc123"
})

datasets_upload / datasets_list

Secure data handling

Upload a dataset and list what you have uploaded. Data is encrypted in transit. 

// Upload returns UUID + key
datasets_upload({
  prefix: "analysis"
})

path_resolver / path_tool_run

Pick the path, then run it

path_resolver executes nothing and says how your question will be answered. path_tool_run then runs a library tool on your data and returns a report link, with the method named. 

// Returns report URL
path_resolver({ objective: "...", dataset_ref: "..." })
path_tool_run( /* path_input from the resolver */ )

reports_list / reports_view

Query past analyses

Search across all past reports with semantic queries. "What do we know about churn?" finds relevant insights from any analysis. 

// Semantic search
reports_list({
  semantic_query: "profit margins"
})

Methods the Library Covers

A sample of the statistical methods in the tool library, reached through the same interface. The list is illustrative, not complete; the docs list what is current.

Regression
  • Linear Regression
  • Ridge / Lasso
  • Elastic Net
  • Logistic
Time Series
  • ARIMA
  • Prophet
  • VAR
  • Exponential Smoothing
ML & Classification
  • Random Forest
  • XGBoost
  • K-Means Clustering
  • Decision Trees
Hypothesis Testing
  • T-Test
  • ANOVA
  • Chi-Square
  • Bayesian A/B
Customer Analytics
  • RFM Segmentation
  • CLV (BG/NBD)
  • Churn Prediction
  • Cohort Analysis
Causal Inference
  • Causal Impact
  • Diff-in-Diff
  • Propensity Matching
  • Synthetic Control
Economics
  • Price Elasticity
  • Demand Forecasting
  • Market Basket
  • Association Rules
Survival & Other
  • Cox Regression
  • Kaplan-Meier
  • PCA
  • Correlation
View All Tools

How It Works

The MCP protocol enables your agent to discover and execute tools on your behalf.

1

Agent Gets Tools

MCP protocol exposes available tools to your agent

2

User Asks Question

"Analyze sales trends for Q4"

3

Agent Selects Tool

Discovers and calls appropriate analysis

4

Results Returned

Report URL + insights for the user

Example Agent Conversation (illustrative sketch, not a recorded run)
>
Can you analyze this sales data and forecast next month's revenue?
*
datasets_upload {"prefix": "sales"}
returns a dataset reference
*
path_resolver {"objective": "forecast revenue", "dataset_ref": "..."}
returns the path it picked, why, and the input to run it
*
path_tool_run {path_input}
returns a report link with the method named
*
The forecast, its range and the method used are in the report linked above. (The agent writes this summary from the report; it is not a fixed template.)

Want to see what comes back? Open a real delivered report, an RFM customer segmentation on a public sample dataset, with its method and numbers stated.

How Your Data Is Handled

What we can state plainly.

API Key Authentication

Secure key-based access to every tool. Keys are managed from your account page.

Encrypted Data Storage

Uploaded data is encrypted in transit. Details are in the docs.

Isolated Execution

You reach only your own datasets and reports.

Power Your Agent with Analytics

One config. A library of statistical tools. Real analysis, callable from an MCP-capable AI.

Try a Free Answer View Documentation
WHERE NEXT
the buildInside the build pipeline →The stages that write, check, and deliver each analysis. why rWhy R →Why the engine runs R and what that buys your results. the shelfBrowse the analysis library →Prebuilt analyses: filter by industry, question, or method.

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.

ds@mcpanalytics.aimcpanalytics.ai

Try it on your own data

30 days free.

Answers, analytics and custom analytics. No card, no call, nothing to install.

30 days free. →
mcpanalytics.ai

Send us my data →