database-snowflake

name	database-snowflake
description	Snowflake data warehouse queries and schema inspection. Use when running SQL queries against Snowflake, listing tables, or inspecting schemas.
allowed-tools	Bash(python *)

Snowflake Data Warehouse

Authentication

IMPORTANT: Credentials are injected automatically by a proxy layer. Do NOT check for SNOWFLAKE_PASSWORD in environment variables - it won't be visible to you. Just run the scripts directly; authentication is handled transparently.

Configuration environment variables you CAN check (non-secret):

• SNOWFLAKE_ACCOUNT - Snowflake account identifier
• SNOWFLAKE_WAREHOUSE - Compute warehouse
• SNOWFLAKE_DATABASE - Default database
• SNOWFLAKE_SCHEMA - Default schema

---

MANDATORY: Schema-First Investigation

Always get the schema before writing queries.

GET SCHEMA / LIST TABLES → DESCRIBE TABLE → EXECUTE QUERY

Available Scripts

All scripts are in .claude/skills/database-snowflake/scripts/

get_schema.py - Get Database Schema (START HERE)

python .claude/skills/database-snowflake/scripts/get_schema.py

list_tables.py - List Tables

python .claude/skills/database-snowflake/scripts/list_tables.py [--database DB] [--schema SCHEMA]

describe_table.py - Table Column Details

python .claude/skills/database-snowflake/scripts/describe_table.py --table TABLE_NAME [--database DB] [--schema SCHEMA]

execute_query.py - Run SQL Queries

python .claude/skills/database-snowflake/scripts/execute_query.py --query "SELECT * FROM fact_incident ORDER BY started_at DESC LIMIT 10" [--limit 100]

---

Investigation Workflow

Incident Data Analysis

1. get_schema.py (understand available tables)
2. execute_query.py --query "SELECT * FROM fact_incident WHERE sev = 'SEV-1' ORDER BY started_at DESC LIMIT 10"
3. execute_query.py --query "SELECT c.customer_name, ic.estimated_arr_at_risk_usd FROM fact_incident_customer_impact ic JOIN dim_customer c ON ic.customer_id = c.customer_id WHERE ic.incident_id = '<id>'"