---
title: "Atlan MCP Server | Enterprise Context for Every AI Tool"
url: "https://atlan.com/mcp/"
description: "Atlan MCP delivers certified business context - definitions, lineage, access rules, and institutional knowledge - to every AI tool and agent your team uses. Cursor, Snowflake Cortex, Claude, n8n, and more."
keywords: "MCP server, model context protocol, Atlan MCP, AI agents, data catalog, business context, AI governance"
---

> Atlan is hosting Context Conference, bringing together the leaders and builders at the frontier of giving AI the context it needs to understand their business. It runs online on October 28, 2026, from 11:00 AM to 2:00 PM ET. Atlan co-founder Prukalpa Sankar opens and closes the day. Leaders from AstraZeneca, BNY and Verizon share why they invest in context and what they get from it. Registrants get early access to The AI Context Gap, a new study from MIT Technology Review Insights. Register: https://atlan.com/context-conference/

**Atlan MCP Server: every tool your team uses, every agent you deploy, one context layer.** Atlan MCP delivers certified business context (definitions, lineage, access rules, and institutional knowledge) to every AI tool and agent your team already uses, without extra configuration. Remote MCP is hosted and managed by Atlan; your AI tool connects via OAuth with the same login you use in Atlan.

[Setup MCP Live](https://atlan.com/build-with-atlan-mcp/) / [Documentation](https://docs.atlan.com/product/capabilities/atlan-ai/how-tos/atlan-mcp-overview)

Works with: n8n, Windsurf, Snowflake, Microsoft Copilot Studio, Databricks, Glean, Google Agent Development Kit, Cursor, Claude, Claude Code, ChatGPT, Gemini, and VS Code.

## Set up MCP in 3 steps

Every tool ends with the same step 3: **authenticate via OAuth** with your Atlan account; your existing roles and data permissions carry over automatically.

| Tool | Step 1 | Step 2 |
|---|---|---|
| Cursor | Open the Atlan plugin on the [Cursor marketplace](https://cursor.com/marketplace/atlan) | Click Install. The Atlan MCP server is added automatically: no config files, no JSON to edit |
| Claude | Open the Atlan plugin in the [Claude plugin directory](https://claude.ai/directory/plugins/atlan%40knowledge-work-plugins) | Click Install to add Atlan to your Claude integrations. No manual MCP config |
| Claude Code | Follow the [Claude Code setup guide](https://docs.atlan.com/product/capabilities/atlan-ai/how-tos/claude-code-plugin-remote-mcp) | Run the install steps; the Atlan MCP server registers with Claude Code automatically |
| ChatGPT | Settings -> Integrations -> Connect apps -> Add MCP connector ([guide](https://docs.atlan.com/product/capabilities/atlan-ai/how-tos/chatgpt-remote-mcp)) | Paste the endpoint `https://mcp.atlan.com/mcp` |
| Gemini | Gemini Advanced or AI Studio -> Integrations -> Add tool -> MCP server ([guide](https://docs.atlan.com/product/capabilities/atlan-ai/how-tos/gemini-remote-mcp)) | Add the endpoint `https://mcp.atlan.com/mcp` |
| VS Code | Open the Atlan extension on the [VS Code Marketplace](https://marketplace.visualstudio.com/items?itemName=atlan.atlan) | Click Install; the server registers with VS Code automatically |
| Cortex Code (Snowflake Cortex Copilot) | Follow the [setup guide](https://docs.atlan.com/product/capabilities/atlan-ai/how-tos/snowflake-cortex-remote-mcp) | Add `https://mcp.atlan.com/mcp` to your Snowflake configuration |
| Genie (Databricks) | Follow the [setup guide](https://docs.atlan.com/product/capabilities/atlan-ai/how-tos/databricks-remote-mcp) | Add `https://mcp.atlan.com/mcp` to your Databricks configuration |

## What your agents can do when they know your business

Atlan MCP connects to the tools your team already uses. What changes is what your agents can answer, and how accurately.

- **AI/BI for data teams.** Ask a data question in plain English; get an answer your data team would sign off on. AI-BI tools fail when they guess at metric definitions. Atlan MCP delivers certified table descriptions, business vocabulary, and access context to every data tool, so AI analysts answer like your best human analysts. Tools: [Snowflake Cortex](https://docs.atlan.com/product/capabilities/atlan-ai/how-tos/snowflake-cortex-remote-mcp), [Databricks Genie](https://marketplace.databricks.com/details/6c60f803-0b80-4bc3-8964-6ec7be190b73/Atlan_Atlan-MCP-Enterprise-Context-Layer), [Gemini](https://docs.atlan.com/product/capabilities/atlan-ai/how-tos/gemini-remote-mcp), [ChatGPT](https://docs.atlan.com/product/capabilities/atlan-ai/how-tos/chatgpt-remote-mcp).
- **IDE for engineering teams.** Debug faster, understand unfamiliar code, know what breaks before you change it. Atlan MCP puts lineage, ownership, and schema context directly in the editor. Tools: [Cursor](https://cursor.com/marketplace/atlan), [VS Code](https://marketplace.visualstudio.com/items?itemName=atlan.atlan), [Windsurf](https://docs.atlan.com/product/capabilities/atlan-ai/how-tos/windsurf-remote-mcp), [Claude Code](https://docs.atlan.com/product/capabilities/atlan-ai/how-tos/claude-code-plugin-remote-mcp).
- **Agentic tools for AI teams.** Agents that run on your context, not training data. Production agents fail quietly when their context goes stale; Atlan MCP feeds every running agent the same certified context your data team maintains, updated automatically and access-controlled by default. Tools: [Google ADK](https://docs.atlan.com/product/capabilities/atlan-ai/how-tos/google-adk-remote-mcp), [n8n](https://docs.atlan.com/product/capabilities/atlan-ai/how-tos/n8n-remote-mcp), [Microsoft Copilot Studio](https://docs.atlan.com/product/capabilities/atlan-ai/how-tos/copilot-studio-remote-mcp), [Glean Agents](https://docs.atlan.com/product/capabilities/atlan-ai/how-tos/glean-remote-mcp).
- **Copilots for everyone.** One context layer for every person and every copilot they use. Answers resolve against your certified definitions, approved metrics, and actual business logic, not what the model predicts your business looks like. Tools: [Claude Desktop](https://docs.atlan.com/product/capabilities/atlan-ai/how-tos/claude-desktop-remote-mcp), ChatGPT, Google Gemini, Glean.

## What production MCP looks like at the enterprises building on it

"Atlan captures Workday's shared language to be leveraged by AI via its MCP server. As part of Atlan's AI labs, we're co-building the semantic layer that AI needs."
 - Joe DosSantos, VP Enterprise Data & Analytics, Workday

"Atlan is much more than a catalog of catalogs. It's more of a context operating system - with an MCP server delivering context to AI models across everything we do."
 - Sridher Arumugham, Chief Data & Analytics Officer, DigiKey

## What teams are building with Atlan MCP

These teams didn't wait for a prescribed workflow. They saw what a governed context layer accessible from any tool, agent, or IDE made possible, and built something with it.

**Building: from lookup tool to building block.** The teams that got the most from Atlan MCP wired it into an agent, a pipeline, or a GitHub workflow.

- "We built our semantic model in Atlan's Context Studio, and Atlan's Remote MCP is how we ground the AI analyst in that context at runtime. The same source of truth, accessed wherever the work happens." - Jessie Buelteman, Senior Director of Data Engineering, Elastic
- "We built a skill in Claude that connects to Atlan via MCP, generates accurate descriptions at scale, and writes them back into Atlan without anyone opening the UI. What would have taken months now runs in hours." - Head of Data, Insurance
- "We built an internal agent that connects to Atlan via MCP and uses our Confluence documentation as context to generate descriptions for tables and columns at scale." - Deputy CTO, Automotive Marketplace
- "Schema Sentinel calls Atlan's MCP on every schema-change PR, checks the lineage, and posts a comment telling the engineer exactly what their change will break - before they merge." - Director of Engineering, Healthcare Data

**Productivity: the answer is already in Atlan; MCP puts it one prompt away**, in the IDE, the chat window, wherever the work happens.

- "We're seeing a lot of value already - here's the definition of certain things, here's how to retrieve it from Atlan. The MCP is a big focus for us right now." - Jon Chen, Director of Enterprise AI/ML Solutions, Workday
- "Having the Atlan MCP connected to Cursor is super useful - the AI writing the code is now aware of the documentation, the context, and the lineage surrounding the model. The data models come out significantly better." - Data Engineer, Technology & SaaS
- "With new data requests, I feed that to Claude and say, go talk to Atlan, go figure out how much of this data we have and where it is." - Director of Enterprise Data & Analytics, Cybersecurity

**Momentum: people start telling their colleagues about it.**

- "I did 3,000 descriptions and 92 READMEs in the last 30 days." - Analytics Manager, Retail & Apparel
- "That was a month project that took an hour." - Director of Enterprise Data & Analytics, Cybersecurity
- "Usage has gone 3x - and I think that is all enabled via the MCP." - Director of Analytics, EdTech
- "Our CTO always talks about force multiplying - taking something and really blowing it out. This is a tool that is doing that right now." - Data Governance Architect, Insurance
- "Atlan is like the compass or directory - where to find the data, which is the relevant source, what other teams use." - Product Marketing Manager, HR Tech

## Go deeper

- [Set Up Atlan MCP in Your Environment](https://docs.atlan.com/product/capabilities/atlan-ai/how-tos/remote-mcp-overview): remote MCP configuration, supported tools, and how context is delivered to your agents.
- [What the Enterprise Context Layer Actually Is](https://atlan.com/know/enterprise-context-layer/): 53+ resources on the infrastructure between your data estate and your AI agents.
- [Build with Atlan MCP](https://atlan.com/build-with-atlan-mcp/): a 45-minute live session for Atlan customers to set up Atlan MCP in the AI tools your team already uses.
- [Talk-to-Data That Actually Works](https://atlan.com/context-engineering-studio/): why AI-BI tools succeed or stall at the context layer, and how teams from DigiKey to Workday connect AI-BI deployments to certified business definitions.

## FAQ

**What is the enterprise context layer?**
The infrastructure between your data estate and your AI tools. It holds your certified definitions, ownership records, lineage traces, access rules, and quality scores: the institutional knowledge that tells an AI tool what a metric means in your organization, not what it might mean in general. Atlan builds this layer by connecting to 400+ source systems and continuously reconciling context across them. When a connected AI tool asks about a table or metric, it gets your org's certified answer, not a trained approximation.

**How is the context layer different from a data catalog or RAG?**
A data catalog stores documentation about your data (descriptions, ownership, tags). A context layer actively delivers that context to connected AI tools at query time, in a format they can use. RAG retrieves documents that might be relevant; the context layer delivers certified, structured context (specific definitions, lineage traces, access controls) from a governed source. RAG-grounded answers are only as good as the retrieved document; context-layer-grounded answers reflect your actual business logic, maintained and certified by the people who own the data.

**What does the context Atlan MCP delivers actually contain?**
Column-level lineage (where data came from and what it feeds downstream), certified metric definitions (the definition your team approved), business glossary terms (what your org means by "revenue," "churn," or "active user"), data quality scores, ownership records, and access control rules (what the requesting user is allowed to see). An agent requesting context about a table gets all of this in a single structured response.

**Can Atlan MCP write back to the catalog?**
Yes. Atlan MCP exposes 29 tools across 8 categories. Read tools (semantic_search, get_assets, traverse_lineage, resolve_metadata, and others) deliver catalog context at query time. Write operations let agents update descriptions, apply or remove classification tags with optional downstream propagation through lineage, set certification status, create glossary terms, and manage announcements from within the AI tool conversation. Every change is marked "Updated using Atlan AI" in the audit trail. Admin tools for data quality rules and custom metadata schemas require Atlan admin permissions.

**Does Atlan MCP send my underlying data to any LLM?**
No. Atlan MCP only delivers metadata: descriptions, lineage, ownership records, glossary terms, certification status, classification tags, and quality scores. Row-level data does not pass through MCP. Metadata is retrieved in real time from Atlan's REST APIs with no intermediate caching, and neither your metadata nor your prompts are used by Atlan or its AI providers to train or fine-tune foundation models. All tool calls are logged as structured audit events with sensitive fields masked before storage.

**How does authentication work?**
Two options. OAuth (recommended for individuals) runs every call as the signed-in user, so their Atlan roles, personas, and access rules apply automatically. API keys, generated from Admin Settings -> API Keys/Tokens, are for service accounts and automation workflows like n8n and Windsurf. API key tokens scoped to multiple personas are rejected; service identities must operate within a single permission scope. Each Atlan tenant has a dedicated MCP server instance with no shared state between tenants.

**Do I need to host or run a server?**
No. Atlan hosts and manages the MCP server for every tenant at `https://your-tenant.atlan.com/mcp`. Most tools (Cursor, VS Code, Claude Code, Snowflake Cortex) connect directly via HTTP. Tools without native HTTP transport (Claude Desktop and Google ADK) use the open-source mcp-remote npm package as a local bridge to the same hosted endpoint; this requires Node.js 18+ installed locally.

**How does the context layer stay current when my data estate changes?**
Atlan's Context Agents continuously read from connected systems (query history, lineage signals, column usage, BI semantic models, documentation) and update the context layer. Because MCP retrieves metadata in real time with no intermediate caching, any enrichment is available to every connected AI tool on the next query. No refresh cycle, no stale cache.

**Can multiple AI tools and agents share the same context layer simultaneously?**
Yes. Atlan exposes context through MCP, SQL, API, and Graph at the same time, so Cursor, a Snowflake Cortex agent, a custom Google ADK agent, and a Glean-powered copilot can all read from the same certified source. When the context layer updates, all of them get the update, which avoids context sprawl: Cortex and your ADK agent stay in sync on the same version of "revenue."

**What is the difference between local MCP and remote MCP?**
Local MCP runs a context server on the developer's machine, so each user manages their own connection and credentials. Remote MCP, Atlan's recommended deployment, runs the server centrally, so every developer, agent, and tool connects to the same governed, current context layer without per-user setup, and context updates reach all connected tools immediately.