Snowflake Basic

@Snowflake Basic show me all tables in the ANALYTICS schema of our PROD database, including row counts

The PROD.ANALYTICS schema contains 12 tables. The largest is CUSTOMER_EVENTS (847M rows, 124GB), followed by PAGE_VIEWS (612M rows, 89GB) and USER_SESSIONS (203M rows, 31GB). Smaller tables include DIM_PRODUCTS (45K rows), DIM_REGIONS (127 rows), and DAILY_AGGREGATES (8.2M rows). All tables were last modified within the past 24 hours except DIM_REGIONS (last updated 6 days ago).

This example demonstrates the MCP's ability to surface table metadata quickly without writing SQL. Requires OAuth access to the specified database and schema. Useful for schema exploration before writing queries, but row counts and sizes are approximate snapshots—not real-time.

@Snowflake Basic run a query to find the top 5 products by revenue in Q1 2024 from SALES.FACT_ORDERS

Query executed successfully in 2.3 seconds. Top 5 products by Q1 2024 revenue: (1) Enterprise License Annual - $4.2M, (2) Professional Services Bundle - $3.8M, (3) Premium Support Contract - $2.1M, (4) Data Integration Add-on - $1.9M, (5) Advanced Analytics Module - $1.7M. Total Q1 revenue across all products: $18.4M. Query scanned 1.2M rows and consumed 0.8 credits.

This write-through query capability lets you execute arbitrary SQL without leaving the conversation. The MCP automatically sets database and schema context before running. Be mindful of query cost—complex queries on large tables consume Snowflake credits. Results are returned inline, not written to a destination table.

@Snowflake Basic describe the CUSTOMER_EVENTS table structure and show me sample values from the EVENT_TYPE column

CUSTOMER_EVENTS has 14 columns: EVENT_ID (NUMBER, primary key), USER_ID (VARCHAR, not null), EVENT_TYPE (VARCHAR, not null), EVENT_TIMESTAMP (TIMESTAMP_NTZ), SESSION_ID (VARCHAR), PAGE_URL (VARCHAR), REFERRER (VARCHAR), DEVICE_TYPE (VARCHAR), and 6 additional metadata columns. Sample EVENT_TYPE values: 'page_view' (68% of rows), 'button_click' (18%), 'form_submit' (7%), 'video_play' (4%), 'download' (3%). The column is indexed and frequently used in WHERE clauses.

Combining schema inspection with column sampling gives you both structure and content context in one prompt. The 'explore columns' tool pulls distinct values with frequency estimates—helpful for understanding categorical data before writing filters. Requires read access to the table; sampling is limited to avoid expensive scans.

When Snowflake Basic beats building custom dashboards for one-off questions

A 5-person ops team fields weekly questions from leadership about customer cohorts, churn windows, or product usage spikes. Building a Looker dashboard for each question takes days and clutters the BI tool. Snowflake Basic wins here: the team runs ad-hoc SQL in Switchy, explores column samples to confirm data shape, then shares the result as a thread artifact. The 6-tool scope is enough for exploratory work—describe table shows schema, run query executes the SQL, show databases navigates the warehouse. The trade-off: if your queries need complex joins across 8+ tables or you're running the same report daily, build the dashboard instead. For one-off questions that change every sprint, Snowflake Basic keeps the team moving without BI overhead.

Onboarding engineers to your warehouse without Slack-bombing the data team

A startup hires its third backend engineer who needs to understand the customer events table before shipping a feature. Normally this means pinging the data lead in Slack five times or reading stale Notion docs. With Snowflake Basic, the engineer opens Switchy, runs show databases and show schemas to map the warehouse, then describe table on the events schema to see column types and nullability. Explore columns pulls sample values to confirm what 'event_type' actually contains. The OAuth2 flow respects existing Snowflake permissions, so the engineer sees only what they're allowed to query. This works until your warehouse hits 50+ schemas—at that point, navigation gets noisy and you need a proper catalog tool. For teams under 15 people with a handful of core tables, Snowflake Basic turns schema discovery into a 10-minute self-service task.

When CS needs to verify account data without waiting on engineering tickets

A 3-person customer success team gets an enterprise client asking why their usage dashboard shows zero API calls last Tuesday. Engineering is in a sprint and the ticket sits for two days. Snowflake Basic lets CS run the spot-check themselves: show tables in the usage schema, describe table on api_logs to confirm the date column format, then run query filtering by client ID and date range. The result confirms the data is there—the dashboard ETL lagged. CS closes the thread in 15 minutes instead of escalating. The boundary: if your CS team needs this more than twice a week, build them a proper internal tool with guardrails. For occasional data verification where the query is simple and the schema is stable, Snowflake Basic keeps CS unblocked without adding engineering load.