Default MCP catalog overview
The default catalog provides tools that allow AI applications and agents to discover, describe, and use GIS capabilities in your ArcGIS Enterprise deployment.
The built-in tools cover four capability areas:
Content and data—Search the portal for items, get information about items and layers, and query records.
Mapping—Export static map images from map services to visualize data, query results, or an area of interest.
Location services—Find locations from addresses or points of interest, find addresses from coordinates, and calculate routes.
Geoprocessing—Get geoprocessing task definitions and check the status of asynchronous jobs.
The default catalog also includes custom MCP tools that your organization adds. For details, see Additional tools in the default catalog.
Available built-in tools
The following tools are included with MCP Services for ArcGIS Enterprise in all supported ArcGIS Enterprise versions. Get GP Task Definition and Get GP Task Job Status are always included in the default catalog. A custom catalog includes them only when it contains a custom MCP tool from a web tool.
| Tool alias | Tool name | What it does |
|---|---|---|
| Search Portal Content | search_portal_content |
Search your organization's portal for items, such as web maps, service items, and apps. Supports structured filters and raw passthrough queries. |
| Describe Item | describe_item |
Get information about a portal item using its item ID. Returns the service definition for service items and the item data in JSON format for other items. |
| Describe Layer | describe_layer |
Get information about a layer or table in a feature service or map service, including fields and example values. Use this information to prepare a query. |
| Query Data | query_data |
Query a feature layer or table using SQL WHERE clauses, spatial filters, pagination, sorting, and statistical aggregations. |
| Get Map Image | get_map_image |
Export a map image from a map service. Supports settings for map extent, image size, output format, layer visibility, attribute filters, and time extent. |
| Find Address Candidates | find_address_candidates |
Find candidate locations for an address, postal code, or point of interest using a configured geocoding service. |
| Reverse Geocode | reverse_geocode |
Find the nearest street address or point of interest for a longitude and latitude coordinate. |
| Solve Route | solve_route |
Calculate a route between a start and end location, with optional intermediate stops and a travel mode, such as driving, walking, or trucking. |
| Get GP Task Definition | get_gp_task_definition |
Get a geoprocessing task's description and its input and output schemas to understand its requirements before running it. |
| Get GP Task Job Status | get_gp_task_job_status |
Check the current status of an asynchronous geoprocessing job using its job ID. |
| Get Search Portal Content Passthrough Instructions | get_search_portal_content_passthrough_instructions |
Get guidance for building a raw passthrough search query when the structured filters in Search Portal Content cannot satisfy the request. |
Note:
Access to the underlying content, data, and services is determined by the permissions of the user or credential used for the connection.
Additional tools in the default catalog
The default catalog also includes custom MCP tools: one for each web tool item tagged for MCP, and any Java-based tools registered with ArcGIS Enterprise. These tools are specific to your deployment, so they are not listed in the table above. For more information, see Expose a web tool as a custom MCP tool and Extend MCP for ArcGIS Enterprise.
The default catalog can include up to 20 web tools and up to 50 Java-based custom MCP tools. If more than 20 web tool items are tagged for MCP, only the 20 most recently modified items are included.
Tool name conflicts
Every tool in a catalog is identified by its name, so two tools with the same name conflict. A conflict can occur between built-in tools, web tools, and Java-based custom MCP tools. Because a web tool's name is based on its service name, a conflict can also occur by chance.
Java-based tools—A Java-based tool with the same name as an existing Java-based tool is rejected when you register it.
Web tools—A web tool with a duplicate name is not rejected when it is published or tagged for MCP. Instead, only one of the conflicting tools appears in the catalog.
In the default catalog, conflicts are resolved in the following order of precedence:
Built-in tools
Java-based custom MCP tools
Web tools. If two web tools have the same name, the most recently modified web tool item is used.
For example, a web tool whose service is named query_data is hidden by the built-in Query Data tool.
In a custom catalog, the tool that is listed first in the catalog item's data is used.
Caution:
Use a unique name for each tool across built-in tools, web tools, and Java-based tools. Rename or remove duplicates before you register a Java-based tool or tag a web tool for MCP.
How tools are documented
Each built-in tool reference page includes the following information:
Tool name—The canonical identifier that MCP clients use to call the tool, such as
search_portal_content.Tool alias—The display name, such as Search Portal Content.
Objective—A brief summary of the tool's purpose.
Usage guidance—An explanation of how the tool works, when to use it, and any usage considerations.
Sample prompts—Example natural-language requests that show how an AI application can use the tool. Get GP Task Definition and Get GP Task Job Status do not include sample prompts because they are called as part of running a web tool.