Skip to main content
Community Hub

Docs-over-MCP tools

TL;DR

Reference for all seven tools exposed by Atlan docs over MCP: what each tool does, its inputs, and when to use it.

Atlan docs over MCP exposes seven tools. Each tool is scoped to a specific type of documentation retrieval. Agents load only what they need, not the entire documentation set.

search_docs​

Search Atlan documentation by keyword or concept. Returns ranked results with titles, paths, URLs, and excerpts. Call this first to find relevant pages, then use get_page to retrieve full content.

querystring
Required

The search query: a concept, feature name, or question.

Example:how to set up Snowflake connector

limitnumber
Optional

Number of results to return. Minimum 1, maximum 20.

Example:5

Examples

  • search_docs("lineage configuration dbt"): finds pages about configuring lineage for the dbt connector, returns paths and excerpts to narrow down which page to fetch next
  • search_docs("Persona access control data product"): surfaces governance pages about Personas, Policies, and how they interact with Data Products
  • search_docs("crawler fails authentication error"): finds troubleshooting pages for crawler authentication issues across connectors
  • search_docs("what metadata does Atlan crawl from Tableau"): returns the Tableau reference pages that list what Atlan extracts

get_page​

Fetch the full Markdown content of a documentation page by its path. Large pages are split into 8,000-character chunks. If the response indicates more chunks are available, call again with chunk incremented.

pathstring
Required

Page path from search_docs results.

Example:/apps/connectors/data-warehouses/snowflake/how-tos/set-up-snowflake

chunknumber
Optional

Chunk index, 0-based. Omit or pass 0 for the first part.

Example:0

Examples

  • get_page("/apps/connectors/data-warehouses/snowflake/how-tos/set-up-snowflake"): full Snowflake setup guide with user creation SQL, required grants, and TCPS configuration
  • get_page("/apps/connectors/data-warehouses/snowflake/how-tos/crawl-snowflake"): Snowflake crawler configuration with extraction methods, connection fields, and schedule options
  • get_page("/product/capabilities/governance/access-control/how-tos/create-a-persona"): step-by-step guide to creating a Persona, setting scope, and attaching Policies
  • get_page("/product/capabilities/lineage/how-tos/configure-query-lineage"): query lineage setup guide with SQL activity mining and miner configuration
  • get_page("/apps/connectors/data-warehouses/snowflake/how-tos/set-up-snowflake", chunk=1): second chunk when the first response says more parts are available

get_feature_guide​

Fetch the full index for a connector or feature area. Returns the table of contents for that area with sections and links to specific pages, so the agent can navigate to exactly what it needs.

slugstring
Required

Connector or feature slug matching the llms hierarchy.

Example:connectors/snowflake

chunknumber
Optional

Chunk index, 0-based. Use when the guide index is large.

Example:0

Examples

  • get_feature_guide("connectors/snowflake"): Snowflake index with Required permissions, Setup options, Crawl and mine, Reference, and Troubleshooting sections, each linking to specific pages
  • get_feature_guide("connectors/dbt"): dbt connector index with authentication setup, project configuration, lineage generation, and job run crawling
  • get_feature_guide("connectors/tableau"): Tableau index with site configuration, workbook crawling, and embedded datasource extraction
  • get_feature_guide("connectors/google-bigquery"): BigQuery index with service account setup, dataset permissions, and column-level lineage
  • get_feature_guide("governance/access-control"): Personas and Policies feature guide with sections on creating Personas, attaching Policies, and masking rules
  • get_feature_guide("governance/data-quality"): Data Quality Studio guide with Snowflake DQ setup, rules, scheduling, and monitoring
  • get_feature_guide("catalog/lineage"): Lineage feature guide with query lineage, ETL lineage, manual lineage, and lineage propagation

get_preflight_requirements​

Fetch the required permissions, roles, and network access needed before setting up a specific connector. Returns the preflight checks page directly, with no need to navigate the full connector guide first.

connectorstring
Required

Connector slug.

Example:snowflake

Examples

  • get_preflight_requirements("snowflake"): Snowflake user creation SQL (CREATE USER, GRANT CREATE SESSION), required grants (SELECT_CATALOG_ROLE, DBA_TABLES), optional TCPS wallet setup
  • get_preflight_requirements("google-bigquery"): BigQuery service account creation, required IAM roles (bigquery.dataViewer, bigquery.jobUser), project-level permissions
  • get_preflight_requirements("dbt"): dbt Cloud API token requirements, account ID location, job permissions needed
  • get_preflight_requirements("tableau"): Tableau site role requirements, personal access token setup, REST API access
  • get_preflight_requirements("oracle"): Oracle dedicated user creation, SELECT_CATALOG_ROLE, DBA view grants, TCPS wallet for encrypted connections
  • get_preflight_requirements("microsoft-sql-server"): SQL Server login creation, VIEW DEFINITION, VIEW DATABASE STATE grants, firewall rules

get_changelog​

Fetch the Atlan product changelog: product changes, new features, integrations, and platform updates organized by month. Use for questions like "what changed in Atlan recently?" or "what integrations were added this year?" The changelog is split into 8,000-character chunks, most recent first. If the response indicates more chunks are available, call again with chunk incremented.

chunknumber
Optional

Chunk index, 0-based. Omit or pass 0 for the most recent entries.

Example:0

Examples

  • get_changelog(): the most recent months of product changes, new connectors, and platform updates
  • get_changelog(chunk=1): the next block of older entries when the first response says more chunks are available
  • get_changelog(), then get_changelog(chunk=1) and onward until the month heading appears: answers "what's new in March 2026?" without loading the whole changelog into context

get_concept​

Fetch the precise Atlan definition and usage context for an Atlan-specific term. Uses Atlan's own vocabulary, not generic equivalents.

termstring
Required

Atlan vocabulary term.

Example:Persona

Examples

  • get_concept("Persona"): a named access profile that controls what data a group of users can see and interact with; not a "role" or "permission group"
  • get_concept("Policy"): a rule attached to a Persona that grants specific permissions on specific assets or domains; not a "permission" or "access rule"
  • get_concept("Lineage"): Atlan distinguishes query lineage (SQL activity mining via the miner) from ETL lineage (ingested by a pipeline connector); not a generic data flow diagram
  • get_concept("Domain"): a business-owned grouping of data assets like "Marketing" or "Finance"; not a folder, tag, or schema
  • get_concept("Classification"): a tag applied to sensitive data for governance enforcement (for example, PII, Confidential) that propagates downstream through lineage by default
  • get_concept("Glossary Term"): a business definition linked to one or more assets; not a generic tag or label
  • get_concept("Asset"): any data object in Atlan (table, column, dashboard, pipeline, model, or file); don't substitute "resource" or "object"

get_concept normalizes plurals and casing, so get_concept("Policies") and get_concept("policy") both return the Policy definition.

When the term isn't one it knows. get_concept answers from a fixed vocabulary plus the concept pages in the llms hierarchy. A term outside both returns no definition, which is a normal empty result rather than an error. The response names the terms it defines and the concept pages it matches, then tells you to call search_docs with the same term. Follow that pointer instead of retrying get_concept with a rephrasing.

explain_page​

Generate a plain-English summary of a documentation page. Unlike the other tools, this one calls a language model to write the summary, so the output is generated from the page rather than quoted from it. Use get_page when you need the exact content, links, or steps.

pathstring
Required

Doc page path, from search_docs results.

Example:/apps/connectors/data-warehouses/snowflake/how-tos/set-up-snowflake

modestring
Optional

written (default) returns a 3 to 5 sentence readable summary. spoken returns a 3 sentence summary written to be read aloud, with no markdown or punctuation that reads badly as speech.

Example:written

Examples

  • explain_page("/apps/connectors/data-warehouses/snowflake/how-tos/set-up-snowflake"): a short summary of what the Snowflake setup guide helps you do and what you can do after reading it
  • explain_page("/product/capabilities/governance/access-control/how-tos/create-a-persona", mode="spoken"): three conversational sentences about creating a Persona, suitable for a voice assistant
  • explain_page("/product/capabilities/lineage/concepts/what-is-lineage"): a plain-language overview of what lineage is before deciding which page to read in full

If the path doesn't match a page, the tool returns an error suggesting search_docs to find the correct path.

See also​