Snowflake MCP. Run complex data queries without switching consoles.
Gemini Works with every AI agent you already use
…and any MCP-compatible client
Just plug in your AI agents and start using Vinkius.
Snowflake MCP Server connects your AI agent directly to Snowflake for data cloud operations. It lets you run complex SQL queries, browse full database schemas, and manage resources—all through conversation.
You can execute `execute_sql` statements, check metadata with `describe_table`, and monitor query status without leaving the chat window.
What your AI agents can do
Cancel sql
Stops an active SQL query that is running too long or consuming excessive resources.
Describe table
Retrieves the column names and data types (schema) for a specified table.
Execute sql
Runs a complete SQL statement against Snowflake, returning results or a job handle.
Run any valid SQL statement against your Snowflake environment using execute_sql.
Browse the data hierarchy by listing databases (list_databases), schemas (list_schemas), and tables (list_tables).
Get a precise breakdown of column names and types for any given table using describe_table.
View, list, and monitor your available data warehouses (list_warehouses) to control cost and performance.
Track the real-time status of running queries using get_statement_status, or halt runaway jobs with cancel_sql.
Ask AI about this MCP
Supported MCP Clients
OAuth 2.0 Compatible Gemini Waiting for input…
Snowflake MCP Server: 11 Tools for Data Management
These tools allow your agent full operational control over your data cloud. You can list everything from databases to users, execute queries, and manage compute resources.
Make your AI actually useful.
Add this MCP to Claude, Cursor, or Windsurf and your AI stops guessing. It gets real tools to look things up, take action, and handle the stuff you keep doing by hand.
Start using Snowflake on Vinkius019dd164cancel sql
Stops an active SQL query that is running too long or consuming excessive resources.
019dd164describe table
Retrieves the column names and data types (schema) for a specified table.
019dd164execute sql
Runs a complete SQL statement against Snowflake, returning results or a job handle.
019dd164get session context
Provides the current working environment details, including database and schema context.
019dd164get statement status
Checks if a specific SQL statement is currently running or has completed successfully.
019dd164list databases
Lists every database accessible within your Snowflake account.
019dd164list roles
Returns a list of security roles defined in the system.
019dd164list schemas
Shows all schemas contained within a given database.
019dd164list tables
Lists tables that reside within a specific schema or database.
019dd164list users
Retrieves a list of user accounts in the Snowflake environment.
019dd164list warehouses
Lists and summarizes all available compute warehouses used for query execution.
Choose How to Get Started
Build a custom MCP for your own tools, or connect a ready-made integration from our catalog.
Build Your Own
Turn any API into an MCP. Import a spec, define Agent Skills, or deploy with MCPFusion.
- Import from OpenAPI, Swagger, or YAML specs
- Create Agent Skills with progressive disclosure
- Deploy to edge with MCPFusion framework
- Built in DLP, auth, and compliance on every call
- Real time usage dashboard and cost metering
- Publish to catalog or keep private
Make Your AI Do More
Start with Snowflake, then connect any of our 4,900+ other servers whenever your AI needs more. One click, no limits.
- Use this MCP plus 4,900+ others, all in one place
- Add new capabilities to your AI anytime you want
- Every connection is secured and compliant automatically
- Track usage and costs across all your servers
- Works with Claude, ChatGPT, Cursor, and more
- New servers added to the catalog every week
Independent Platform Disclaimer: Vinkius is an independent platform and is not affiliated with, endorsed by, sponsored by, verified by, or otherwise authorized by Snowflake. All third-party trademarks, logos, and brand names are the property of their respective owners. Their use on this website is strictly for informational purposes to identify service compatibility and interoperability.
VINKIUS INFRASTRUCTURE
Cloud Hosted
Managed infra
V8 Isolated
Sandboxed per request
Zero-Trust Proxy
No stored credentials
DLP Enforced
Policy on every call
GDPR Compliant
EU data residency
Token Compression
~60% cost reduction
Works with Claude, ChatGPT, Cursor, and more
The Model Context Protocol standardizes how applications expose capabilities to LLMs. Instead of operating in isolation, your AI gains direct access to external platforms, live data, and real-world actions through secure, standardized connections.
This server provides 11 capabilities that interface natively with Claude, ChatGPT, Cursor, and any MCP client. No middleware. No custom integration required.
Trying to map out your data structure feels like clicking through five different consoles.
Today, figuring out where a piece of data lives is tedious. You jump from the database console to the schema viewer, then into the table metadata page just to confirm column names. It's a mess of tabs, copy-pasting identifiers, and waiting for slow loading screens just to write one SELECT statement.
With this MCP server, you talk to your agent about it. You ask: 'Where is customer revenue data?' The agent uses `list_databases`, then narrows the scope using `list_schemas` until they pinpoint the exact table. The whole discovery process—the tedious part—gets condensed into a natural conversation.
Snowflake MCP Server: Get answers directly from your data warehouse.
Before this, running an ad-hoc query meant opening the SQL editor, manually typing `USE DATABASE X; USE SCHEMA Y;` just to ensure context, and hoping you didn't typo a table name. It was high effort for basic retrieval.
Now, you simply ask your agent using natural language commands. The server handles the full execution flow: it confirms your current environment with `get_session_context`, runs the query via `execute_sql`, and gives you the results—all while keeping track of potential issues like running jobs that need to be terminated with `cancel_sql`.
What you can do with this MCP connector
Yo, listen up. This Snowflake MCP Server hooks your AI agent right into your data cloud for serious operations. You're not just running queries; you're managing the whole damn stack—all through conversation.
When you connect your agent to this server, it gives you direct access to run complex SQL and handle everything from mapping out your databases to monitoring compute costs.
To start figuring things out, your agent can first check what’s running in your environment with get_session_context, which hands you the current database and schema details. If you gotta scope out the whole system, it'll list every accessible data repository using list_databases. From there, if you want to see all the organizational containers inside a specific database, use list_schemas.
Need to know what tables are sitting in that schema? Run list_tables, which shows you exactly which tables reside there. If you need to dig deeper into one of those tables—say, checking out column names and data types—you'll call describe_table. This gives you the precise schema breakdown before you write a single line of SQL.
For writing queries, you use execute_sql to run any valid statement against Snowflake; it’ll spit back results or give you a job handle if the query is big. If that query starts hogging resources or just runs too long, you've got ways to manage it: first, check its progress with get_statement_status, and if it won't quit, hit the brakes with cancel_sql.
Beyond data querying, this thing lets you handle system oversight. It can list all user accounts in your environment using list_users or review who has permissions by calling list_roles. To keep an eye on costs and performance, it lists every compute warehouse available via list_warehouses, letting you control what’s running.
It's basically a full operational interface. You can map the entire data hierarchy by listing databases (list_databases), schemas (list_schemas), and tables (list_tables); then, check column types with describe_table; run any SQL statement using execute_sql while tracking its status with get_statement_status, or stopping runaway jobs instantly with cancel_sql. You can also view your current setup with get_session_context, and see the full roster of system users (list_users), roles (list_roles), and compute resources (list_warehouses).
019dd164-6be4-728e-8895-18b173f86d07 
How Snowflake MCP Works
- 1 Subscribe to the server and provide your Snowflake Account Identifier and necessary access token (JWT or OAuth2).
- 2 Tell your agent what you want. Example: 'List all tables in the PROD database.' The agent uses
list_tables. - 3 After identifying a target table, ask the agent to use
describe_table. Once the schema is confirmed, ask it to run the final query usingexecute_sql.
The bottom line is that you get full data cloud control—from resource listing to executing complex queries—without leaving your chat interface.
Who Is Snowflake MCP For?
Anyone who has to interact with a large, multi-schema data warehouse without the luxury of dedicated GUI tools. This is for the BI Engineer who spends too much time clicking through console dashboards; it's for the Data Scientist needing immediate schema validation before every query.
Retrieving data across multiple schemas and ensuring they pull the right columns by first running describe_table.
Automating metadata retrieval—listing all relevant databases (list_databases) or monitoring resource usage via list_warehouses to keep reports running fast.
Rapidly prototyping queries on large datasets, needing the ability to check context with get_session_context and analyze results immediately.
What Changes When You Connect
- Stop guessing column names. Use
describe_tableto get the exact schema details before you write a single line of code, preventing runtime errors and wasted computation. - Manage query execution risk. If a report hits a bottleneck, check status with
get_statement_status, or kill it instantly usingcancel_sqlinstead of waiting minutes for a timeout. - See your entire data map at a glance. Tools like
list_databases,list_schemas, andlist_tableslet you navigate the whole structure right inside the chat, eliminating manual console navigation. - Control cost and performance by knowing what's running. Use
list_warehousesto check active compute resources before kicking off a large job, ensuring your query runs on the correct size warehouse. - Work with context awareness.
get_session_contexttells you exactly which database and schema the agent is operating under, making multi-step queries reliable.
Real-World Use Cases
The Schema Discovery Problem
A data scientist needs to join customer records but doesn't know if 'address' lives in the CUSTOMERS schema or the SALES schema. Instead of running multiple guesses, they ask their agent: 'List all schemas in the PROD database.' The agent uses list_schemas, and once the correct area is identified, they use describe_table to confirm the required columns before running the join via execute_sql.
Debugging a Slow Report
A BI engineer runs a massive report that hangs. They don't know if it's stuck or just processing. First, they use get_statement_status to check the progress. If it's clearly stalled, they hit 'cancel' using cancel_sql. This saves them from waiting hours for nothing.
Auditing Data Access
A compliance officer needs to know which groups can see sensitive user data. They use the agent to list roles (list_roles) and then check who belongs to those roles using list_users. This provides an immediate audit trail of access permissions.
Resource Allocation Check
A development team needs to run a complex ETL process. Before running the job, they ask: 'What warehouses are active?' The agent uses list_warehouses, allowing them to ensure the correct compute resource is available and properly configured for the workload.
The Tradeoffs
Directly executing complex joins
The user tries to write a massive, multi-join query using execute_sql immediately. This often fails because they missed a required schema or column name, resulting in an ambiguous error and wasting time.
→
Never jump straight to the final query. First, use list_databases, then list_schemas, followed by describe_table on every table involved. Confirming the full path minimizes errors before calling execute_sql.
Ignoring resource limits
The user runs a huge query during peak hours and hits a performance ceiling, or worse, spikes the cost without realizing it.
→
Check resources first. Use list_warehouses to see available compute units. If you need guaranteed speed, check availability via the tool before running large jobs.
Assuming context
The user runs a query and assumes it hits the right data set because they're in the 'sales' project. They forget to specify the schema or database.
→
Always verify your location first. Run get_session_context before executing any critical SQL statement, even if you think the context is already set.
When It Fits, When It Doesn't
Use this server when your task involves deep discovery and multi-step validation: You need to go from 'I want X' -> 'Where does X live?' (using list_databases, list_schemas) -> 'What columns are available there?' (describe_table) -> 'Run the final query.'
Don't use this if your goal is simple, point-in-time data retrieval where a dedicated BI GUI handles all the schema mapping for you. If you just need to view a chart based on pre-validated fields, an existing dashboard tool is faster. But when you need to build a new report or validate schemas yourself, this server gives you the full operational control necessary. Remember that running execute_sql can be costly; always use list_warehouses first.
Common Questions About Snowflake MCP
How do I check if a long-running SQL statement is still active using get_statement_status? +
You pass the Query ID (or Job ID) to the agent. The tool checks Snowflake and returns the current status, telling you if it's 'RUNNING,' 'SUCCEEDED,' or 'FAILED.' This prevents you from waiting on a dead query.
Is there a way to browse all available tables without listing everything? +
Not directly. You must use list_databases first, then select the correct database, followed by list_schemas, and finally list_tables. It's a hierarchical drill-down process.
Can I see what columns are in a table before querying it using describe_table? +
Yes. Running describe_table immediately gives you the full schema—all column names and their data types—so you can write your query accurately without guessing.
What if my query is running too long and I need to stop it? +
Use the cancel_sql tool. The agent sends a cancellation request for the specific job ID, stopping compute usage immediately and freeing up resources.
How do I check which databases are available using list_databases? +
It immediately returns a list of all accessible databases. This is your starting point for data exploration, letting you see the highest level of structure in your Snowflake account without needing to guess names.
What information does get_session_context provide about my current connection? +
get_session_context gives you metadata about your current session. This includes details like the active user, assigned warehouse, and time zone settings—essential for confirming that your agent is running in the right environment.
How do I see all available compute warehouses using list_warehouses? +
list_warehouses lists every compute resource you can use. You check this to manage costs and performance, letting you route different types of queries (like ETL vs. reporting) to dedicated, isolated environments.
What does list_roles show me regarding security permissions? +
list_roles provides a clear inventory of all defined security roles in your account. This is critical for data governance because it shows exactly what permission sets exist before you try to access restricted data.
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.
Use it with your favorite AI tools
Connect this server to Cursor, Claude, VS Code, and more.