Atlan context for AI

Your AI tools work better with Atlan when they can access the actual documentation—not cached training data. Connect Claude, Cursor, Copilot, or any MCP-compatible tool using Atlan docs over MCP once and it retrieves the right page automatically, grounded in the real docs rather than guesswork. Atlan docs over MCP is separate from Atlan MCP, Atlan's metadata connectivity product.

⚡ Connect AI context

Set up Atlan docs over MCP in your AI tool once. From then on, ask questions in plain language—the tool finds and fetches the right page automatically.

Set up docs over MCP→See all tools→

$ claude mcp add --transport http atlan-docs https://docs.atlan.com/mcp

What your AI can do

🔍

Search documentation

Find the right page by concept, connector name, or question — ranked results with paths and excerpts

🔌

Get connector setup guides

Load the full index for any connector — Snowflake, dbt, Tableau, BigQuery — with all setup, crawl, and reference sections

🔐

Check required permissions

Retrieve the exact SQL grants, IAM roles, or API tokens needed before setting up a connector

💡

Look up Atlan concepts

Get precise definitions for Persona, Policy, Domain, Classification, Lineage — from Atlan's own vocabulary

📄

Fetch full page content

Retrieve the complete Markdown of any doc page, chunked for large pages so nothing is missed

📋

Copy or view as Markdown

Copy any page to your clipboard or open it as raw Markdown in a new tab — no MCP server required

How it works

Atlan docs over MCP organizes documentation as a four-level hierarchy—a master index, area indexes, connector or feature indexes, and individual page files. The tools navigate this hierarchy efficiently, loading only what's relevant rather than the entire documentation set.

1

Connect your AI tool

Set up Atlan docs over MCP once. Works with Claude Code, Claude Desktop, Cursor, and Copilot.

→

2

Ask in natural language

Your AI tool picks the right tool — search, fetch, or concept lookup — based on what you asked.

→

3

Get accurate answers

Responses are grounded in the live documentation — the same source as the docs site.

AI agent patterns

Beyond consuming the docs, Atlan is the governed context and metadata layer behind your AI agents. These how-tos cover common patterns for building, serving, and monitoring agent context with Atlan as the enabling layer:

• Build a context layer: your first single-domain build for one team and one AI client.
• Build an enterprise context layer: a multi-domain, governed, compliance-ready program.
• Set up a memory layer for AI agents: long-term, governed agent memory using context repositories and MCP.
• Monitor your AI agents with Atlan: data-side observability that pairs with your LLM observability tool.
• Write an AGENTS.md file: instruct AI coding agents and point them at Atlan context.