# Snowflake MCP for AI Agents AI Agent Connect > Snowflake MCP lets you run SQL queries, manage your database structure, and monitor your warehouse resources using natural language. It gives your AI agent direct access to your data cloud, making it easy to explore tables, check session details, and track query performance without leaving your chat interface. ## Overview - **Category:** industry-titans - **Price:** Free - **Endpoint:** https://edge.vinkius.com/vk_preview_Rp0KQ9PtMeNiS3B9qz1nNvDYeWgTbTyxK57hAOY5/ai-agent-connect - **Tags:** sql-query, data-warehousing, cloud-data, database-schema, data-analysis ## Description You can stop jumping between your SQL editor and your chat window. Instead of manually typing out complex JOINs or hunting through nested schemas to find the right table, you just tell your agent what data you need. It handles the heavy lifting of navigating your data cloud environment. You can ask it to find specific tables, describe what columns they contain, or check which warehouses are currently running. It even handles the boring stuff like monitoring query statuses or figuring out your current session context. Because this is hosted on the Vinkius catalog, you get a reliable way to bridge the gap between your data warehouse and your everyday AI workflow. It turns your data cloud into a conversational interface where you can pull insights, manage permissions, and keep an eye on costs in one place. This means you spend less time wrestling with syntax and more time actually looking at the numbers. You can verify data types, check user roles, and see what's happening with your compute resources without ever opening a separate tab. It's about making your data accessible through a simple conversation, letting you move from a question to a result without the usual manual steps. ## Tools ### get_session_context Retrieves the current session context from your account. It helps you understand the specific environment and permissions your agent is currently using. ### get_statement_status Checks whether a specific SQL statement is still running or has finished. This is useful for monitoring long-running queries. ### list_databases Shows all the databases you have permission to see in your account. It provides a quick overview of your available data sources. ### list_roles Lists the security roles available within your Snowflake environment. Use this to audit permissions and manage access. ### list_schemas Shows all the schemas located within a specific database. It helps you navigate your data architecture quickly. ### list_tables Lists all the tables found in a particular schema or database. This makes it easy to find the right data for your queries. ### list_users Lists all the users currently registered in your Snowflake instance. It is useful for administrative oversight and user management. ### list_warehouses Shows all the compute warehouses available for your queries. Use this to monitor your compute resources and costs. ### cancel_sql Stops a SQL statement that is currently running in your warehouse. This prevents long queries from wasting your credits. ### describe_table Provides the schema details and column information for a specific table. It lets you see what data is available before you write a query. ### execute_sql Runs a SQL statement and returns the first partition of results or a handle for longer queries. This is the primary way to pull data into your chat. ## Prompt Examples **Prompt:** ``` What tables do we have in the retail_data schema? ``` **Response:** ``` I've looked into the `retail_data` schema for you. Here are the tables I found: * `daily_sales` * `inventory_levels` * `store_locations` * `product_catalog` Would you like me to describe any of these for you? ``` **Prompt:** ``` Show me all the active warehouses and their current status. ``` **Response:** ``` Here is the current status of your Snowflake warehouses: | Warehouse | Status | Role | | :--- | :--- | :--- | | `COMPUTE_WH` | Running | `TRANS_ROLE` | | `LOAD_WH` | Suspended | `ETL_ROLE` | | `DEV_WH` | Running | `DEV_ROLE` | Your `COMPUTE_WH` is currently active and ready for queries. ``` **Prompt:** ``` Run a query to find the top 5 customers by total spend in 2024. ``` **Response:** ``` I've executed that query for you on the `ANALYTICS_WH`. **Top 5 Customers (2024):** 1. **Global Retail Inc.** - $450,200 2. **Tech Solutions** - $320,150 3. **Urban Living** - $210,400 4. **Prime Goods** - $185,000 5. **Eco Friendly Co.** - $142,300 Total rows scanned: 12,400. ``` ## Capabilities ### Run SQL queries Execute statements and get back results for immediate analysis. ### Browse database schemas See your databases, schemas, and tables in a clear list. ### Monitor warehouse usage Check which compute resources are active and see their status. ### Describe table structures Get details on columns and data types without manual lookups. ### Manage user roles View and list the security roles across your environment. ### Track query progress Check if a long-running statement is still active or finished. ### Cancel active queries Stop a running SQL statement if it's taking too long. ## Use Cases ### Finding sales data An analyst asks for 'top 10 products by revenue' and the agent uses `execute_sql` to pull the numbers from the correct schema. This saves them from having to write and test the SQL manually before seeing the results. ### Auditing permissions A security admin asks the agent to 'list all roles' and then checks specific permissions to ensure the production environment is locked down. It provides a quick way to audit access without manual navigation. ### Checking warehouse health A manager asks 'which warehouses are active' and the agent uses `list_warehouses` to show current utilization. This helps keep a pulse on resource usage during peak hours. ### Data exploration A new team member asks 'what tables are in the marketing schema' and the agent uses `list_tables` to give them a starting point. This helps them get oriented in the data cloud faster. ## Benefits - Stop manual schema hunting by using `list_schemas` and `list_tables` to quickly find the data you need without memorizing every table name. This lets you move from a vague question to a specific query in seconds. - Get instant query results with `execute_sql` to see your data immediately in your chat window instead of waiting for a separate tool to load. You can see the numbers right where you're working. - Monitor costs and performance using `list_warehouses` and `get_statement_status` to ensure your compute resources aren't being wasted. You can see exactly what's running and how much it's costing. - Understand your data structure faster with `describe_table` to see exactly what columns are available before you start writing complex queries. It removes the guesswork from your data exploration. - Keep your environment organized by using `list_roles` and `list_users` to audit permissions and security settings through simple chat commands. This makes staying compliant much easier for your team. - Maintain control over long-running jobs with `cancel_sql` to stop queries that are eating up your credits or taking too much time. You stay in control of your warehouse spend at all times. ## How It Works The bottom line is you get a direct conversational bridge to your Snowflake data warehouse. 1. Subscribe to the Snowflake MCP and connect it to your AI client. 2. Provide your Snowflake Account Identifier and a valid JWT or OAuth2 token. 3. Ask your agent to run queries or list tables to start managing your data. ## Frequently Asked Questions **Can the Snowflake MCP run complex SQL queries?** Yes, you can ask your agent to execute specific SQL statements and get back results. It handles the heavy lifting of running the code and presenting the data to you in the chat. **Does the Snowflake MCP let me see my database structure?** It does. You can ask the agent to list your databases, schemas, and tables so you don't have to memorize the names or navigate through menus. **Can I use the Snowflake MCP to monitor my warehouse costs?** You can use it to list your warehouses and check the status of running statements. This helps you keep track of your resource usage in real-time. **Is the Snowflake MCP safe for my data?** Yes, it works with your existing Snowflake credentials like JWT or OAuth2. Your agent only performs the actions you specifically ask it to do. **Can I cancel a query if it's running too long?** Absolutely. You can ask the agent to check the status of a specific statement and then tell it to cancel it if it's not finishing as expected. **What kind of roles can I manage with the Snowflake MCP?** You can ask the agent to list the security roles in your account, which is great for auditing permissions and keeping your environment secure. **How do I find my Snowflake Account Identifier?** Your Account Identifier is the part of your Snowflake URL before '.snowflakecomputing.com'. It typically looks like `xy12345.us-east-2.aws`.