What features does Snowflake MCP server provide?

Q: What features does Snowflake MCP server provide?

Dynamic memo insights resource: memo://insights is a continuously updated memo aggregating discovered data insights; it updates automatically when new insights are appended via the append_insight tool.. Contextual per-table schema resources (prefetch): context://table/{table_name} exposes per-table schema summaries (columns and comments) when prefetching is enabled.. Read-only SQL query tool: read_query executes SELECT queries and returns results as an array of objects.. Write-enabled SQL modification tool: write_query: write_query executes INSERT/UPDATE/DELETE queries and returns the number of affected rows or a confirmation; enabled only with --allow-write.. Table creation tool: create_table: create_table executes CREATE TABLE statements; returns a confirmation; enabled only with --allow-write.. Schema discovery toolset: list_databases, list_schemas, list_tables, and describe_table allow metadata discovery (databases, schemas, tables, and detailed table schemas).. Insight analysis tool: append_insight: append_insight adds new data insights to the memo resource and triggers its update.. Access control and configurability: By default, write operations are disabled and can be enabled with --allow-write; supports exclusion patterns to filter databases, schemas, or tables and optional prefetching of per-table context.

Snowflake MCP Server 2025: Features & Setup Guide

Overview

Snowflake MCP Server is an MCP server implementation that enables LLMs to interact with Snowflake databases. It provides tools to run SQL queries and manage data operations within a controlled environment, exposing data context and schema information as resources. The server offers resources such as memo://insights, a continuously updated memo of discovered data insights that updates automatically when new insights are appended via the append_insight tool; and context://table/{table_name}, which (if prefetch is enabled) exposes per-table schema summaries including columns and comments. The available tools include read_query for executing SELECT queries and returning results as objects; write_query and create_table, both enabled only with --allow-write, for data modification and table creation; and a schema tools suite (list_databases, list_schemas, list_tables, describe_table) for metadata discovery. An analysis tool, append_insight, adds new insights to the memo resource. By default, write operations are disabled; exclusion patterns can be used to filter out databases, schemas, or tables. The README also provides installation and configuration guidance for Claude Desktop via Smithery or UVX, and local usage steps with Snowflake credentials and connection parameters.

Features

Dynamic memo insights resource

memo://insights is a continuously updated memo aggregating discovered data insights; it updates automatically when new insights are appended via the append_insight tool.

Contextual per-table schema resources (prefetch)

context://table/{table_name} exposes per-table schema summaries (columns and comments) when prefetching is enabled.

Read-only SQL query tool

read_query executes SELECT queries and returns results as an array of objects.

Write-enabled SQL modification tool: write_query

write_query executes INSERT/UPDATE/DELETE queries and returns the number of affected rows or a confirmation; enabled only with --allow-write.

Table creation tool: create_table

create_table executes CREATE TABLE statements; returns a confirmation; enabled only with --allow-write.

Schema discovery toolset

list_databases, list_schemas, list_tables, and describe_table allow metadata discovery (databases, schemas, tables, and detailed table schemas).

Insight analysis tool: append_insight

append_insight adds new data insights to the memo resource and triggers its update.

Access control and configurability

By default, write operations are disabled and can be enabled with --allow-write; supports exclusion patterns to filter databases, schemas, or tables and optional prefetching of per-table context.