MCP For Claude Marketplace
Snowflake

Snowflake MCP Connector for Claude

A+

Bring your absolute data cloud into your AI editor. Execute queries, list warehouses, and map complex schemas natively.

7 tools Official Updated Jun 28, 2026 Official Vinkius Partner
More Details Connect to Claude

Connect your Snowflake AI Data Cloud with your AI agent to radically accelerate the way you query large datasets and audit cloud data warehouses. Navigate through deep hierarchical trees of databases, tables, and internal stages natively by chatting with your IDE. Keep your SQL robust by validating commands directly against the live engine.

What you can do

  • Execute Queries in Chat — Tell your bot to execute_sql based on human prompts or test new complex table joins safely right inside Cursor or Claude
  • Map Infrastructures — Quickly retrieve spatial contexts by pulling list_databases, traversing downwards through list_schemas to target specific columns
  • Audit Compute Cost — Keep a firm grip on active clusters running by auditing running instances using list_warehouses
  • Diagnose Operations — Monitor long-tail data workloads or data engineering pipelines using the get_query_status method asynchronously

How it works

  1. Subscribe to this AI integration server
  2. Introduce your explicit Snowflake Account identifier (e.g. abc123.us-east-1)
  3. Inject your Snowflake OAuth token or JWT Token (key pair) authentication string
  4. Ask Claude or Cursor to look into the Sales Database schema

Stop juggling browser instances to paste a quick query in Snowflake Snowsight. Stay strictly inside your local codebase while examining the exact table data types.

Who is this for?

  • Data Engineers — validate that raw datasets correctly land in internal environments (list_stages) straight from your IDE window
  • Analytics Engineers / dbt — generate highly accurate SQL modeling by letting your agent examine the list_tables definitions live
  • Software Architects — write an agentic script pulling raw diagnostic query metrics without downloading hefty SDK kits locally
sql-warehousedata-cloudquery-executionschema-mappingdata-warehousingbig-data

7 tools expose this connector's capabilities to your AI agent.

list_databases

Lists all databases in the Snowflake account

list_schemas

Lists all schemas within a specific database

list_tables

Lists all tables within a specific schema

execute_sql

Prefers read-only statements whenever possible. Executes a SQL query on Snowflake

list_warehouses

Lists all virtual warehouses

list_stages

Lists all internal and external stages

get_query_status

Retrieves the status of an asynchronous query

See how to talk to your AI agent using Snowflake.

List all running virtual warehouses I can access in my Snowflake account.

I've called `list_warehouses` for your targeted Snowflake footprint. **Running Clusters Detected**: 1. `COMPUTE_WH` – Size: X-Small | State: SUSPENDED 2. `ANALYTICS_WH` – Size: Large | State: STARTED 3. `ETL_LOAD_WH` – Size: Medium | State: STARTED Would you like me to run an analytical snippet under `ANALYTICS_WH` now?

Write a query to grab the top 5 most engaged users from our schema and execute it.

No problem! I constructed this query to target exactly that and executed it against the system: `SELECT user_id, email, activity_score FROM prod_schema.users ORDER BY activity_score DESC LIMIT 5;` **Snowflake Output Data**: 1. alice@example.com (Score: 980) 2. john84@test.io (Score: 964) 3. m_reyes@corp.com (Score: 941) [...] Do you want me to format this as a markdown table or save the dataset?

Retrieve the schema mapping for the MASTER_DB. I need to know all nested tables before doing table joints.

Using `list_schemas`, `MASTER_DB` contains main partitions: 1. `PUBLIC` 2. `ANALYTICS_GOLD` Digging into `ANALYTICS_GOLD` using `list_tables`: - `dim_customers` (Type: BASE TABLE) - `dim_geography` (Type: BASE TABLE) - `fct_sales_records` (Type: BASE TABLE) - `vw_live_revenue` (Type: VIEW) I am ready to help build your exact JOINS. For example, `fct_sales_records.customer_id` directly joins on `dim_customers.id`.

Yes. When the AI uses `execute_sql` with something like `SELECT * FROM schema.users LIMIT 10`, the MCP integration parses the exact row outputs. The LLM consumes the tabular data back into context so you can converse naturally about the dataset findings.

Related Connectors

FR

Franchimp MCP

12 tools Official

Access franchise intelligence, lead gen data, and FDD metadata via AI agents with Franchimp.

A+ View details →
JS

Jira Service Management (JSM) MCP

10 tools Official

Manage service desks, customer requests, and queues via Jira JSM API.

A+ View details →
JA

Jasper MCP

20 tools Official

Equip your AI agent with direct access to Jasper — generate marketing copy, manage brand voices, and orchestrate content campaigns without opening the Jasper app.

A+ View details →
LO

Logz.io MCP

31 tools Official

Query logs, manage alerts, and monitor triggered events in Logz.io directly from your AI agent using Elasticsearch DSL.

A+ View details →