Unity Catalog MCP Server
An MCP server that allows LLM agents to seamlessly execute functions within Unity Catalog.
mcp-server-unitycatalog: An Unity Catalog MCP server
Overview
A Model Context Protocol server for Unity Catalog. This server provides Unity Catalog Functions as MCP tools.
Tools
You can use all Unity Catalog Functions registered in Unity Catalog alongside the following predefined Unity Catalog AI tools:
-
uc_list_functions- Lists functions within the specified parent catalog and schema.
- Returns: A list of functions retrieved from Unity Catalog.
-
uc_get_function- Gets a function within a parent catalog and schema.
- Input:
name(string): The name of the function (not fully-qualified).
- Returns: A function details retrieved from Unity Catalog.
-
uc_create_function- Creates a function within a parent catalog and schema. WARNING: This API is experimental and will change in future versions.
- Input:
name(string): The name of the function (not fully-qualified).script(string): The Python script including the function to be registered.
- Returns: A function details created within Unity Catalog.
-
uc_delete_function- Deletes a function within a parent catalog and schema.
- Input:
name(string): The name of the function (not fully-qualified).
- Returns: None.
Installation
Using uv
When using uv no specific installation is needed. We will use
uvx to directly run mcp-server-git.
Configuration
These values can also be set via CLI options or .env environment variables. Required arguments are the Unity Catalog server, catalog, and schema, while the access token and verbosity level are optional. Run uv run mcp-server-unitycatalog --help for more detailed configuration options.
| Argument | Environment Variable | Description | Required/Optional |
|---|---|---|---|
-u, --uc_server | UC_SERVER | The base URL of the Unity Catalog server. | Required |
-c, --uc_catalog | UC_CATALOG | The name of the Unity Catalog catalog. | Required |
-s, --uc_schema | UC_SCHEMA | The name of the schema within a Unity Catalog catalog. | Required |
-t, --uc_token | UC_TOKEN | The access token used to authorize API requests to the Unity Catalog server. | Optional |
-v, --uc_verbosity | UC_VERBOSITY | The verbosity level for logging. Default: warn. | Optional |
-l, --uc_log_directory | UC_LOG_DIRECTORY | The directory where log files will be stored. Default: .mcp_server_unitycatalog. | Optional |
Usage with Claude Desktop or VSCode Cline
Add this to your claude_desktop_config.json (or cline_mcp_settings.json):
Using uv
{
"mcpServers": {
"unitycatalog": {
"command": "uv",
"args": [
"--directory",
"/<path to your local git repository>/mcp-server-unitycatalog",
"run",
"mcp-server-unitycatalog",
"--uc_server",
"<your unity catalog url>",
"--uc_catalog",
"<your catalog name>",
"--uc_schema",
"<your schema name>"
]
}
}
}
Using docker
- Note: replace '/Users/username' with the a path that you want to be accessible by this tool
{
"mcpServers": {
"unitycatalog": {
"command": "docker",
"args": [
"run",
"--rm",
"-i",
"mcp/unitycatalog",
"--uc_server",
"<your unity catalog url>",
"--uc_catalog",
"<your catalog name>",
"--uc_schema",
"<your schema name>"
]
}
}
}
Building
Docker:
docker build -t mcp/unitycatalog .
Future Plans
- Implement support for
list_functions. - Implement support for
get_function. - Implement support for
create_python_function. - Implement support for
execute_function. - Implement support for
delete_function. - Implement semantic catalog explorer tools.
- Add Docker image.
- Implement
use_xxxmethods. In the current implementation,catalogandschemaneed to be defined when starting the server. However, they will be implemented asuse_cataloganduse_schemafunctions, dynamically updating the list of available functions when theuse_xxxis executed.
License
This MCP server is licensed under the MIT License. This means you are free to use, modify, and distribute the software, subject to the terms and conditions of the MIT License. For more details, please see the LICENSE file in the project repository.
Server Terkait
Unofficial Reactome MCP Server
Access Reactome pathway and systems biology data via its live API.
MCP JDBC Server
A lightweight MCP server for any database with a JDBC driver. Built with Quarkus and requires Java 21+.
PostgreSQL Full Access MCP Server
A full-access PostgreSQL server for MCP with read/write capabilities and enhanced schema metadata.
InfluxDB MCP Server
An MCP server for interacting with InfluxDB time-series databases, enabling AI assistants to work with time-series data.
Quanti: connectors MCP
Unify your marketing team around one AI-powered source of truth. Quanti connects your marketing data to your warehouse. Execute SQL queries on BigQuery, explore table schemas, discover pre-built use cases, and analyze performance across Google Analytics, Google Ads, Meta Ads, TikTok, affiliate networks and more. all through natural conversation
FinDataMCP
Provides financial data. Requires external Python dependencies installed with the uv package manager.
PostgreSQL
Provides read-only access to PostgreSQL databases, allowing LLMs to inspect schemas and execute queries.
Neo4j
Neo4j graph database server (schema + read/write-cypher) and separate graph database backed memory
OracleDB MCP Server
An MCP server for interacting with Oracle Database, enabling SQL generation and data retrieval via LLM prompts.
Mantora
Mantora is a local-first MCP observer: a lightweight UI + proxy for inspecting LLM data access (sessions, tool calls, results) with protective defaults.