Actian Data Intelligence Platform MCP Server¶
The Actian Data Intelligence Platform MCP Server uses the open-source Model Context Protocol (MCP) standard to enable AI agents to access contextual metadata from external systems. This protocol provides a standardized way for large language models (LLMs) and automation frameworks to retrieve the context they need to provide accurate and reliable results.
The Actian Data Intelligence Platform MCP Server acts as a secure bridge between the Actian Data Intelligence Platform and AI tools, such as Claude, Cursor, Windsurf, and Microsoft Copilot Studio. With the Actian Data Intelligence Platform MCP Server, you can search and find assets at both the semantic layer and the data model level, using real-time context from Studio and Explorer applications within the Actian Data Intelligence.
Tools¶
The Actian Data Intelligence Platform MCP Server provides tools that enable AI agents to work directly with metadata in the Actian Data Intelligence. These tools supply real-time context to AI environments, making it easier to search, explore, and update metadata without leaving your workflow.
Search and Discovery¶
Search Items¶
The search_items tool is the primary search tool for retrieving catalog items based on filter criteria. Use it to find any type of item in the catalog, including datasets, fields, visualizations, data processes, and more.
| Parameter | Type | Default | Description |
|---|---|---|---|
item_type_code | string | — | Code of the item type (for example, dataset, field, visualization). Use get_metamodel_item_types to discover available item type codes. |
catalog_code | string | — | Code of the catalog to scope results to. Use list_catalogs to discover available catalog codes. |
name | string | "" | Name of the item (partial match). |
description | string | "" | Text in the description (partial match). |
properties | object | {} | Filter by property values. The key is the property code and the value is the value to match (for example, {"$z_schema": "music"}). |
cursor | string | — | Cursor from a previous response's next_cursor. Omit for the first page. |
limit | integer | 20 | Maximum number of items to return (1–100). |
visible_in_explorer | boolean | true | When true, only return items visible in Explorer. Set to false to include Studio-only items. |
Example: Search for datasets named albums in the music schema:
{
"item_type_code": "dataset",
"name": "albums",
"properties": {"$z_schema": "music"}
}
Each result includes the item's key, which can be passed to get_item to read full item details.
Semantic Search Glossary¶
The semantic_search_glossary tool performs an AI-powered semantic search to find glossary terms (business definitions) conceptually related to a given query, based on meaning rather than exact keywords.
Note
This tool is only for finding glossary items (business vocabulary, definitions, and concepts). Do not use it to find datasets, tables, fields, or other technical items.
| Parameter | Type | Default | Description |
|---|---|---|---|
query | string | — | Required. Reference or question for semantic search. |
item_type_code | string | — | Keep only results for this item type (for example, dataset). |
limit | integer | 20 | Maximum number of items to return (1–100). |
visible_in_explorer | boolean | true | When true, only return items visible in Explorer. |
Results are relevance-ranked (most relevant first). There is no pagination. Increase limit to surface more results.
Example: If you search for revenue, the tool returns related terms such as income, sales, gross profit, and billing.
Search Contacts¶
The search_contacts tool discovers contacts (people or teams) by name or email.
| Parameter | Type | Default | Description |
|---|---|---|---|
name | string | "" | Name of the contact (partial match). Matches across first and last names. |
email | string | "" | Email address of the contact (partial match). |
cursor | string | — | Cursor from a previous response's next_cursor. Omit for the first page. |
limit | integer | 20 | Maximum number of items to return (1–100). |
List Catalogs¶
The list_catalogs tool lists all catalogs accessible to the authenticated tenant. Use the returned codes as the catalog_code parameter in search_items to scope results to a specific catalog.
| Parameter | Type | Default | Description |
|---|---|---|---|
offset | integer | 0 | Number of items to skip (zero-based). |
limit | integer | 20 | Maximum number of items to return (1–100). |
Item Details¶
Get Item¶
The get_item tool retrieves all available information about a single item, including its names, descriptions, summary, catalog and source property values, and type-specific information (for example, quality level and category for a dataset, data type and key constraints for a field, or version for a visualization).
| Parameter | Type | Default | Description |
|---|---|---|---|
item_ref | string | — | Required. Reference to the item. Prefer the key returned by search_items. An item ID can also be used, but resolving an item by name is slower. |
Get Dataset Data Model¶
The get_dataset_data_model tool retrieves the relational data model for a specified dataset, including its fields (columns) with their names and types, relationships to other datasets (such as foreign keys and joins), and connected datasets.
| Parameter | Type | Default | Description |
|---|---|---|---|
dataset_ref | string | — | Required. Reference to the dataset (name, ID, or URI such as zee://connection/schema/table). |
Example: If you query the orders table, the tool can show its relationships to the customers, products, and shipping tables.
Get Item Lineage¶
The get_item_lineage tool retrieves the lineage graph for a catalog item, including its upstream producers and downstream consumers. It returns both nodes and directed edges.
| Parameter | Type | Default | Description |
|---|---|---|---|
item_ref | string | — | Required. Name, ID, or URI of the item. |
direction | string | both | upstream for items it is produced from, downstream for items produced from it, or both for both directions. |
depth | string | next | next for directly adjacent items only, all for the entire reachable graph. |
For a field, the tool returns the entire field-to-field lineage graph. In this case, direction and depth parameters do not apply.
Get Item Audit Events¶
The get_item_audit_events tool retrieves the audit and activity history for a catalog item, including who changed what and when.
| Parameter | Type | Default | Description |
|---|---|---|---|
item_ref | string | — | Required. Name, ID, or URI of the item. |
since | string | — | Include only events at or after this ISO-8601 date and time. |
until | string | — | Include only events at or before this ISO-8601 date and time. |
offset | integer | 0 | Number of items to skip (zero-based). |
limit | integer | 20 | Maximum number of items to return (1–100). |
Glossary¶
Find Glossary Definition¶
The find_glossary_definition tool retrieves the stored definitions of a glossary term from the data catalog.
| Parameter | Type | Default | Description |
|---|---|---|---|
item_name | string | — | Required. Name of the glossary item. |
offset | integer | 0 | Number of items to skip (zero-based). |
limit | integer | 20 | Maximum number of items to return (1–100). |
visible_in_explorer | boolean | true | When true, only return items visible in Explorer. |
Example: If you query Customer Lifetime Value, the tool returns the exact definition text stored in the glossary.
Get Glossary Implementations¶
The get_glossary_implementations tool retrieves all data assets (such as datasets) that are linked as implementations of a specified glossary term using the Implements predicate.
| Parameter | Type | Default | Description |
|---|---|---|---|
glossary_item_ref | string | — | Required. Name, ID, or URI of the glossary item. |
item_type_code | string | — | Keep only implementations of this item type (for example, dataset). |
cursor | string | — | Cursor from a previous response's next_cursor. Omit for the first page. |
limit | integer | 20 | Maximum number of items to return (1–100). |
visible_in_explorer | boolean | true | When true, only return items visible in Explorer. |
Example: If you search for Customer, the tool returns datasets that implement this concept, such as customer_profile_table, customer_transactions, and customer_support_tickets.
Metamodel¶
The metamodel defines the structure and configuration of the data catalog itself, not the data it contains.
Get Metamodel Item Types¶
The get_metamodel_item_types tool retrieves a list of all possible item types defined in the catalog's metamodel (for example, dataset, field, and visualization).
| Parameter | Type | Default | Description |
|---|---|---|---|
name_contains | string | — | Keep only item types whose name contains this text (case-insensitive). |
code | string | — | Keep only the item type with this exact code (for example, dataset). |
layer | string | — | Keep only item types in this layer: physical (technical/data assets) or semantic (business/glossary concepts). |
offset | integer | 0 | Number of items to skip (zero-based). |
limit | integer | 20 | Maximum number of items to return (1–100). |
Get Metamodel Properties¶
The get_metamodel_properties tool retrieves a list of all possible properties defined in the catalog's metamodel, which can include the common properties of connectors such as $z_schema and $z_table.
| Parameter | Type | Default | Description |
|---|---|---|---|
name_contains | string | — | Keep only properties whose name contains this text (case-insensitive). |
code | string | — | Keep only the property with this exact code (for example, $z_schema). |
property_type | string | — | Keep only properties of the specified data type: Date, Enum, MultivaluedEnum, Number, RichText, ShortText, Tag, Uri, or Url. |
item_type_code | string | — | Keep only properties applicable to this item type. |
offset | integer | 0 | Number of items to skip (zero-based). |
limit | integer | 20 | Maximum number of items to return (1–100). |
List Responsibilities¶
The list_responsibilities tool retrieves the list of all responsibility roles defined in the catalog, such as Owner and Steward.
| Parameter | Type | Default | Description |
|---|---|---|---|
offset | integer | 0 | Number of items to skip (zero-based). |
limit | integer | 20 | Maximum number of items to return (1–100). |
Curation¶
List Suggestions¶
The list_suggestions tool lists curation suggestions, including pending, accepted, or rejected recommended changes (such as proposed descriptions and property values) created by automation or users for steward review.
| Parameter | Type | Default | Description |
|---|---|---|---|
item_ref | string | — | Name, ID, or URI of a catalog item to scope suggestions to. Omit to list all suggestions. |
status | string | — | Filter by suggestion status: Pending, Accepted, or Rejected. Omit for all statuses. |
cursor | string | — | Cursor from a previous response's next_cursor. Omit for the first page. |
limit | integer | 20 | Maximum number of items to return (1–100). |
Write Tools¶
Update Item Description¶
The update_item_description tool sets or clears the catalog description of an item.
| Parameter | Type | Default | Description |
|---|---|---|---|
item_ref | string | — | Required. Reference to the item. Prefer the key returned by search_items. |
content | string | — | Description text to set. Omit together with content_type to clear the description. |
content_type | string | — | Format of content: RAW for plain text and HTML for rich text (sanitized). Required when content is provided. |
Update Item Summary¶
The update_item_summary tool sets or regenerates the summary displayed for an item in search results.
| Parameter | Type | Default | Description |
|---|---|---|---|
item_ref | string | — | Required. Reference to the item. |
summary | string | — | Summary text to set (maximum 1000 characters). Omit to regenerate the summary from the item's current description. |
Update Item Property¶
The update_item_property tool sets or clears the value of a custom property on an item. Use get_metamodel_properties to look up a property's code and type before setting a value.
| Parameter | Type | Default | Description |
|---|---|---|---|
item_ref | string | — | Required. Reference to the item. |
property_ref | string | — | Required. Reference to the property: its ID or code, as returned by get_metamodel_properties. |
value | string, number, or array | — | Value to set, shaped according to the property's type. Omit to clear the property. |
The expected shape of value depends on the property's type:
| Property Type | Value Shape |
|---|---|
ShortText, Enum | string |
Number | number |
MultivaluedEnum, Tag | array of strings (for example, ["hello", "world"]) |
Date | ISO 8601 date or date-time string |
Url | URL or absolute path string |
Properties of type Uri or RichText cannot be updated through this tool. Read-only properties (exclusively from a source scanner) cannot be set or cleared.
Prompts¶
The Actian Data Intelligence Platform MCP Server provides pre-built prompts that help AI agents accomplish common tasks. These prompts guide the agent through multi-step workflows using the available tools.
| Prompt | Description | Parameters |
|---|---|---|
explore_schema | Explore the structure and contents of a database schema. | schema_name — Name of the schema to explore. |
understand_dataset | Understand a dataset's structure, fields, and relationships. | dataset_reference — Reference to the dataset (name, ID, or URI). |
find_business_term | Find a business term, concept, or definition in the glossary. | concept — Business term, concept, or question to search for. |
find_items | Find items of a specific type that match a concept. | item_type — The type of items to filter by. concept — Concept included in the items to search for. |
Pagination¶
Tools that return lists use one of two pagination styles:
-
Offset-based pagination: Used by
find_glossary_definition,list_catalogs,get_item_audit_events,get_metamodel_item_types,get_metamodel_properties, andlist_responsibilities.Parameter or Field Description offsetNumber of items to skip (zero-based). Default: 0.limitMaximum items per page (1–100). Default: 20.next_offsetPass this value as offsetto retrieve the next page.nullwhen on the last page. -
Cursor-based pagination: Used by
search_items,get_glossary_implementations,search_contacts,semantic_search_glossary, andlist_suggestions.Parameter or Field Description cursorOpaque cursor from a previous response's next_cursor. Omit for the first page.limitMaximum items per page (1–100). Default: 20.next_cursorPass this value as cursorto retrieve the next page.nullwhen on the last page.
All paginated responses include items, total_count, limit, and has_more. When has_more is true, request the next page using next_offset or next_cursor.
Testing with MCP Inspector¶
The MCP Inspector is a tool for testing and debugging MCP servers. Use it to verify your connection to the Actian Data Intelligence Platform MCP Server and explore the available tools.
Setup¶
-
Start the MCP Inspector:
npx @modelcontextprotocol/inspector -
Configure the connection:
- Transport type:
Streamable HTTP - URL:
https://<your-tenant>.zeenea.app/api/mcp-server/mcp - Request Timeout:
30000 - See Remote Authentication Setup for authentication details.
- Transport type:
-
Click Connect.