The server itself exposes 41 tools, and some of them can write to your Atlan account. Behind TrustGate an agent only sees the tools you allow, requests and responses are inspected, and every call is logged. Anything outside the allowlist is denied before it reaches Atlan.

How is this different from connecting Atlan MCP directly?

Connected directly, an agent sees every tool the server exposes and holds the credential itself. Through TrustGate you choose which tools each agent gets, and every call is logged against the agent that made it.

What if the Atlan MCP API changes?

We track the Atlan MCP schema, so the tool catalogue stays current. Tools you have already allowed keep working, and anything new stays denied until you enable it.

Can I give different agents different Atlan tools?

Yes. Access is set per agent: one can get a handful of read tools, another the full set. You grant the whole MCP or single tools within it.

Where can I see what my agents did in Atlan?

Every call is logged against the agent that made it, with the tool it used. Activity can be sent on to your own monitoring.

Do I need to run the Atlan MCP server myself?

No. TrustGate connects to the Atlan MCP server and sits in front of it. Your agent only talks to your TrustGate URL, never to Atlan directly.

Connect Atlan MCP to your AI agents, securely.

One secured endpoint for every Atlan call your AI agents make, scoped per user, inspected, and logged.

Get started for free

Get a demo

Try Atlan MCP now

What your agents can ask once Atlan MCP is connected.

Select the tools your AI agents can call

41 tools from the Atlan MCP schema.

Showing 41 tools

semantic_searchreadSearches the entire Atlan catalog in natural language with intent-aware ranking. It is the same search agent that powers Conversational AI.
askreadHands an open-ended question to Atlan AI, which plans the work, queries the context layer, iterates and returns one grounded answer. A single call typically takes 30 to 120 seconds.
search_assetsreadFinds assets using structured filters: type, certification status, tags, domains, glossary terms and dates, with exclusions, OR matching, sorting and aggregations. Returns 20 results per page by default, up to 100.
get_assetsreadRetrieves complete details for one or more assets by GUID: metadata, owners, tags, custom attributes and related entities such as columns, glossary terms and README. It takes GUIDs from a prior search, not names.
search_atlan_docsreadSearches Atlan's product documentation and returns a generated answer with links to the source pages. For questions about Atlan itself, not for finding data assets.
resolve_metadatareadResolves tenant-level definitions, Atlan tags and custom metadata sets, by name or keyword, returning their exact names and GUIDs before another call uses them.
get_usersreadFinds workspace users by name, email or username, or fetches one by GUID, for owner and steward assignment and for resolving an owner GUID back to a person.
get_groupsreadLists workspace groups and optionally their members, so an agent can resolve group names to actual identities.
describe_asset_typereadLists the valid attributes and relationships for an asset type, so a follow-up get_assets call requests exactly the type-specific fields it needs.
fluent_searchreadRun a structured catalog search written as a FluentSearch expression — the Atlan Python SDK's query builder — parsed and compiled server-side, never executed as arbitrary code.
fluent_search_schemareadLook up the real fields and operators a fluent_search expression can use for a given asset type, discovered live from the tenant's schema.
traverse_lineagereadTraces upstream or downstream lineage for an asset to understand dependencies, data flows and impact. Returns immediate neighbours by default and traverses deeper hops on request.
query_assetsreadRuns read-only SQL against a connected warehouse such as Snowflake, Redshift, BigQuery or Databricks to retrieve real rows. Statements that modify data are rejected, up to 100 rows are returned, and the caller needs query permission on the connection.
update_assetswriteModifies asset metadata such as descriptions, certification status and README content.
update_custom_metadatawriteSets or updates custom metadata attribute values on one or more assets. Partial by default, so unspecified attributes keep their values; full replacement clears them and applies to one asset at a time.
remove_atlan_tagwriteRemoves a specific classification tag from one or more assets.
manage_announcementswriteAdds or removes informational, warning or issue announcements on assets.
add_atlan_tagshigh riskApplies one or more classification tags to one or more assets. Propagation to downstream lineage is on by default, so one call can tag an entire dependency graph; it can be restricted to lineage only, hierarchy only, or turned off.
remove_custom_metadatahigh riskClears all custom metadata values from an asset for a specific metadata set.
create_custom_metadata_sethigh riskDefines one or more custom metadata sets with typed attributes: string, number, boolean, date, enum, users, groups, URL and more. Schema only. Requires admin permissions in Atlan.
add_attributes_to_cm_sethigh riskAdds new typed attributes to an existing custom metadata set. Requires admin permissions in Atlan.
remove_attributes_from_cm_sethigh riskArchives specific attributes on a custom metadata set. Archived attributes are soft-deleted and their values are cleared from every asset. Requires admin permissions in Atlan.
delete_custom_metadata_sethigh riskPermanently deletes an entire custom metadata set and all its attribute definitions. Requires admin permissions in Atlan.
manage_asset_lifecyclehigh riskArchives, restores or purges assets. Archive and restore are reversible; purging is permanent and cannot be undone.
create_glossarieswriteDefines new business glossaries with metadata and descriptions so teams can organize terms by domain or function.
create_glossary_categorieswriteAdds categories and subcategories inside glossaries so business terms are grouped into navigable hierarchies.
create_glossary_termswriteCreates individual business terms with names, definitions and certificate status.
create_domainswriteDefines data domains and subdomains to organize assets and data products by business area and ownership.
create_data_productswriteCreates data products linked to assets within a domain so teams can publish and discover business-ready datasets.
create_dq_ruleswriteDefines rules on critical tables and views to enforce baseline data quality: completeness, uniqueness, validity, freshness, volume, reconciliation and custom SQL checks.
update_dq_ruleswriteAdjusts thresholds, priorities and conditions on existing rules.
schedule_dq_ruleswriteConfigures cron-based schedules for rules on tables and views so checks run automatically alongside ETL workflows.
delete_dq_ruleshigh riskRemoves deprecated or noisy rules, which silently stops a quality check from running.
read_artifactreadBrowses and reads Atlan's artifact catalog of skills, knowledge files and agents. With no arguments it lists everything; with a GUID or name it returns the artifact as markdown with its file content inlined.
search_knowledgereadSearches uploaded knowledge files semantically and returns focused passages with citations: file name, folder, page number and document title. Early preview.
list_knowledgereadLists knowledge folders, or the files inside one folder, to see what context has already been uploaded. Early preview.
read_knowledge_filereadReads the full text of one knowledge file, for summarizing, translating or analyzing a whole document. Early preview.
create_knowledge_folderwriteCreates a folder to organize uploaded documents such as PDFs, Markdown and YAML that agents read for business context. Idempotent. Early preview.
save_knowledge_contentwriteUploads a document as a knowledge file and indexes it in the same call, so it is immediately searchable. Handles text formats directly and binaries such as PDF and DOCX up to 10 MB. Early preview.
get_knowledge_upload_urlwriteReturns a presigned upload URL for large or binary knowledge files, paired with register_knowledge_file. Early preview.
register_knowledge_filewriteRegisters a file in the catalog after a presigned upload completes, making it listable and readable. Early preview.

What Atlan MCP exposes

Atlan MCP groups its 41 tools into 9 categories, each with a suggested starting policy.

All 41 tools as JSON, ready to paste.

Category	Covers	Access	Suggested policy
Search & Discovery	Natural-language semantic search across the catalog, an open-ended agent that plans and iterates before answering, structured asset filters with aggregations, full asset detail by GUID, Atlan product documentation search, tenant-level tag and custom metadata resolution, workspace users and groups, and the valid attributes for an asset type	Read	Read-only. ask is an agent that plans, queries and iterates, so one call takes 30 to 120 seconds and its scope is open-ended. get_users and get_groups resolve real people and group memberships
Knowledge & Artifacts	Browsing and reading Atlan's artifact catalog of skills, knowledge files and agents, semantic search over uploaded documents with citations, listing and reading knowledge files, creating folders, uploading and indexing content, and presigned uploads for large binaries	Read / Write	Mostly early preview and may not be enabled on a tenant. Uploaded documents become agent-readable business context, so what goes in here shapes what every later answer is grounded on
Asset Metadata	Modifying asset descriptions, certification status and README content, setting custom metadata values, applying and removing classification tags, and managing informational, warning and issue announcements	Write	add_atlan_tags propagates to downstream lineage by default, which can be narrowed to lineage only, hierarchy only or turned off. update_custom_metadata is partial by default; full replacement clears unspecified attributes
Custom Metadata	Defining custom metadata sets with typed attributes, adding attributes to an existing set, archiving attributes, clearing a set's values from an asset, and permanently deleting a whole set	Write / Admin	Four of these five require admin permissions in Atlan. Archiving an attribute soft-deletes it and clears its values from every asset, and deleting a set removes all its attribute definitions permanently
Data Quality	Defining rules for completeness, uniqueness, validity, freshness, volume, reconciliation and custom SQL checks on critical tables and views, adjusting thresholds and conditions, cron scheduling, and removing rules	Write	schedule_dq_rules puts checks on a cron alongside ETL, so a rule created here runs unattended afterwards. Deleting a rule silently stops a quality check that something downstream may depend on
Business Glossary	Creating glossaries, categories and subcategories, and individual business terms with definitions and certificate status	Write	Creation only, no update or delete in this surface. Glossary terms are the organization's shared definitions, so new ones carry editorial weight more than operational risk
Lineage & Exploration	Tracing upstream or downstream lineage to understand dependencies, data flows and impact, and running read-only SQL against a connected warehouse to retrieve real rows	Read	query_assets is the one path from catalog metadata to actual records. It rejects statements that modify data, caps at 100 rows and needs query permission on the connection, but the rows themselves are production data
Data Domains	Defining data domains and subdomains by business area and ownership, and creating data products linked to assets within a domain	Write	Domains and data products organize ownership, so creating them asserts who is responsible for what. Worth an approval even though nothing is destroyed
Asset Lifecycle	Archiving, restoring and purging assets to manage the end-of-life state of deprecated or replaced data assets	Write	One tool with three very different outcomes. Archive and restore are reversible; purge is permanent and cannot be undone, so gate on the action argument rather than the tool name

Connect Atlan MCP to your AI agent

Four steps, under ten minutes.

01
Sign in to NeuralTrust
Open the NeuralTrust platform, sign in and go to Getting started, then Connect your tools.
02
Choose your agent
When asked which agent you use, pick Claude Code, Codex, Cursor, Windsurf, Copilot or your own code.
03
Connect Atlan
Find Atlan in the tool list and connect it.
04
Connect your agent
Follow the steps on screen to add the config to your agent, then send the test prompt.

Permissions down to the single tool

You do not allow Atlan MCP or block it. You allow the exact tools each agent can call, one by one.

Atlan MCP	41 TOOLS AVAILABLE
Code review agent	4 tools, read only
Issue triage bot	9 tools
Platform team assistant	all 41 tools

Tool by toolConnecting the server is not the unit of access. Each tool is granted on its own.
Default denyA tool you have not enabled is never exposed to the agent.
Approval on writesTools that change or delete things can require a human to confirm.

Is Atlan MCP secure?

One endpoint instead of raw credentials. Every call is checked against policy and logged.

AI agentsConnect through your MCP client configuration.	NeuralTrustApply controls based on your configuration.	Atlan MCP41 documented tools, each labelled Read, Write or Admin by Atlan. Every tool that changes something returns a preview and waits for approval before writing. Tools are gated per tenant, so a client may list fewer than the full set, and the knowledge tools are in early preview.
Coding assistants	Connection & credentialsper user	Search & Discovery11 tools
Internal copilots	Tool access policiesdeny by default	Knowledge & Artifacts8 tools
Workflow automations	Request & response inspectioninline	Asset Metadata5 tools
Compatible MCP clients	Activity monitoringalerts and logs	All other tools17 tools