---
title: Snowflake
summary: Query data and manage tables in Snowflake.
---

# Snowflake

Query data and manage tables in Snowflake.

## What it does

Connects an agent to Snowflake. Agents can run a query, list databases and schemas, list tables, and describe a table's structure.

## Setup

1. In Snowsight, generate a programmatic access token and copy your account's URL from the account details page.
2. Go to **Settings → Integrations → Snowflake** and enter the account URL, access token, warehouse, and database.
3. Assign the connection to an agent.

If you connected Snowflake before v1.0.75, remove that connection and reconnect with a programmatic access token; the earlier username-and-password setup never actually worked.

## Credentials needed

Your Snowflake account URL, a programmatic access token, warehouse, and database. The token is encrypted at rest and never shown again after saving. Snowflake only issues a token to a user covered by a network policy, and a token expires after 15 days by default, so plan to generate a new one periodically.

## Typical use

An agent that runs a read-only query to answer a question about your own data, or one that describes a table's structure before writing a more complex query against it.

## Related

- [Integrations Guide](/en/docs/guides/integrations-guide): the full catalog and security model these connections share
- [Airtable](/en/docs/integrations/airtable): a lighter-weight alternative for smaller, less structured datasets