RudderStack MCP Server

Control Your Data Pipelines from Your AI Client

https://mcp.rudderstack.com/mcp
What Is It?

The RudderStack MCP (Model Context Protocol) server connects your AI client (Claude Desktop, Cursor, VS Code) directly to your RudderStack workspace. Ask questions, debug pipelines, create transformations, and monitor data flows—all through natural language in your favorite AI tool.

Your AI Client
(Claude/Cursor)
→
RudderStack MCP Server
→
Your RudderStack
Workspace
New to RudderStack? (What the workspace controls)

RudderStack is a warehouse-native customer data infrastructure that lets you collect once, store in your own warehouse, and activate everywhere without surrendering control.

Teams lean on RudderStack to:

  • Collect trustworthy behavioral data via 15+ SDKs and dozens of cloud apps.
  • Unify identities and enforce tracking plans before the data hits your stack.
  • Activate enriched data through Reverse ETL, personalization, and predictive models.
Sources
Web SDKs
JavaScript, AMP, Service Worker
Mobile SDKs
iOS, Android, React Native, Flutter
Server SDKs
Node, Python, Go, .NET, Rust
Cloud App Sources
Shopify, Braze, Iterable, more
→
Data Plane
Event Stream
Workspace-managed routing & transforms
→
Your Warehouse
Your Warehouse
You own & control all data
→
Destinations
200+ Cloud Tools
Analytics & Marketing

Complete data control: Every event lands in your Snowflake, BigQuery, Databricks, or lakehouse environment. RudderStack never persists customer data; you keep full visibility and governance.

Two-layer architecture:

  • Control Plane (Workspace): define sources, destinations, transformations, tracking plans, and permissions. This is also where Event Stream configurations live and what the MCP server exposes for AI assistance.
  • Data Plane: executes the Event Stream. Run it as RudderStack Cloud or self-host the open-source rudderlabs/rudder-server so no data ever traverses RudderStack infrastructure.
Dive Deeper: Introduction to RudderStack →
Key Use Cases

Pipeline Debugging

Need root cause for failing events
  • Monitor live event flow
  • Check success rates by source

Transformation Dev

Mask emails before sending them to the destinations
  • Test with sample events
  • Deploy to production

Data Quality

Audit this week's signup events for duplicate names
  • Find duplicate event names
  • Check tracking plan compliance

Warehouse Ops

Diagnose Mixpanel RETL sync running behind schedule
  • Check sync status
  • Troubleshoot failures

Quick Docs

Find setup docs for the Shopify source
  • Search official docs
  • Access integration guides
How Easy Is It?
1

Add to Config

Add RudderStack MCP to your AI client's configuration file

2

Authenticate

Restart your client and complete the OAuth login flow

3

Start Asking

That's it! Your AI can now access your workspace

{
  "mcpServers": {
    "rudderstack": {
      "command": "npx",
      "args": ["-y", "mcp-remote", "https://mcp.rudderstack.com/mcp"]
    }
  }
}
What You Can Do

Pipeline Config

  • List sources & destinations
  • View configurations
  • Browse integrations

Transformations

  • Create & update
  • Test & deploy
  • View metrics

Event Analysis

  • View volumes
  • Examine schemas
  • Monitor live events

Data Governance

  • Tracking plans
  • Event catalog
  • Audit logs

Warehouse & RETL

  • Sync status
  • View history
  • Troubleshoot failures
Security Built-In

OAuth Authentication

Same permissions as your account

PII Protection

Event payloads auto-hashed

Read-Only Default

Write access limited to transformations

Your Control

Choose your AI model

Get Started

Prerequisites: RudderStack workspace + MCP-compatible AI client

(Claude Desktop, Cursor, VS Code, etc.)

https://mcp.rudderstack.com/mcp

Transform how you work with data pipelines. Instead of switching between UIs—just ask your AI.