Snowflake Object Inventory¶

Purpose: Complete map of every Snowflake database, schema, and table/view in production. Claude Code uses this to avoid naming conflicts and understand the current state of the warehouse.

Last Updated: 2026-03-02 Maintained by: Keep this current when adding new databases, schemas, or tables.

How to Update This File¶

When you create a new Snowflake object (via schemachange migration or dbt), add it here immediately. This prevents duplicate names, aids debugging, and gives Claude Code accurate context for every session.

Databases¶

Database	Client	Environment	Owner	Purpose
`WBP_DB`	Warby Parker	Production	XO_DATA_ENGINEER	Medallion layers: Bronze → Silver → Gold
`WBP_DB_DEV`	Warby Parker	Development	XO_DATA_ENGINEER	Dev/testing (mirrors WBP_DB structure)
`CND_DB`	Condé Nast	Production	XO_DATA_ENGINEER	Medallion layers: Bronze → Silver → Gold
`CND_DB_DEV`	Condé Nast	Development	XO_DATA_ENGINEER	Dev/testing
`CORE_DB`	Shared	Production	XO_DATA_ENGINEER	Shared resources: employee rosters, shared dims
`CORE_DB_DEV`	Shared	Development	XO_DATA_ENGINEER	Dev/testing
`OPERATIONS`	Legacy	Production	XO_DATA_ENGINEER	Legacy reporting DB — being deprecated

WBP_DB (Warby Parker)¶

BRONZE Schema¶

All Bronze tables: all VARCHAR columns, batch_replace loading strategy.

Table	Source System	Load Strategy	DAG	Schemachange Version
`GLADLY_CONTACT_TIMESTAMPS`	Gladly API	batch_replace	warbyparker_gladly_daily	V1.1.0
`GLADLY_CONVERSATION_TIMESTAMPS`	Gladly API	batch_replace	warbyparker_gladly_daily	V1.1.0
`GLADLY_WORK_SESSIONS`	Gladly API	batch_replace	warbyparker_gladly_daily	V1.3.0
`GLADLY_AGENT_DURATIONS`	Gladly API	batch_replace	warbyparker_gladly_daily	V1.4.0
`GSHEETS_CSAT`	Google Sheets	incremental	warbyparker_gsheets_daily	V1.5.0
`GSHEETS_QA`	Google Sheets	incremental	warbyparker_gsheets_daily	V1.5.0

SILVER Schema¶

All Silver tables: dbt dynamic tables, typed columns, deduplicated by RECORD_KEY.

Table	Source Bronze Table	dbt Model	Grain
`CONTACT_TIMESTAMPS`	GLADLY_CONTACT_TIMESTAMPS	`silver/warbyparker/contact_timestamps.sql`	One row per CONTACT_ID
`CONVERSATION_TIMESTAMPS`	GLADLY_CONVERSATION_TIMESTAMPS	`silver/warbyparker/conversation_timestamps.sql`	One row per CONVERSATION_ID
`WORK_SESSIONS`	GLADLY_WORK_SESSIONS	`silver/warbyparker/work_sessions.sql`	One row per AGENT_ID + SESSION_START
`AGENT_DURATIONS`	GLADLY_AGENT_DURATIONS	`silver/warbyparker/agent_durations.sql`	One row per AGENT_ID + STATUS_START
`INBOX_GLOSSARY`	External / Manual	N/A (managed separately)	One row per INBOX_ID

GOLD Schema¶

Object	Type	dbt Model	Source	Grain
`FCT_CONTACTS`	Dynamic Table	`gold/warbyparker/fct_contacts.sql`	SILVER.CONTACT_TIMESTAMPS + CORE_DB.SILVER.ROSTER_WARBYPARKER	One row per CONTACT_ID
`AGG_AGENT_DAILY_CONTACTS`	Dynamic Table	`gold/warbyparker/agg_agent_daily_contacts.sql`	FCT_CONTACTS	One row per (AGENT_ID, CONTACT_DATE)
`RPT_EMAIL_DAILY`	View	`gold/warbyparker/rpt_email_daily.sql`	AGG_AGENT_DAILY_CONTACTS	One row per (AGENT_ID, DATE)
`AI_AGENTS`	Seed Table	`seeds/ai_agents.csv`	Static CSV	One row per agent (AI wave enablement dates)

CND_DB (Condé Nast)¶

BRONZE Schema¶

Table	Source System	Load Strategy	DAG
`SPROUT_MESSAGES`	Sprout Social API	incremental	condenast_sprout_messages_incremental

SILVER Schema¶

Table	Source	dbt Model	Grain
(To be documented as Silver layer is built)

GOLD Schema¶

Object	Type	Notes
(To be documented as Gold layer is built)

CORE_DB (Shared Resources)¶

SILVER Schema¶

Shared resources used by all client Gold layers.

Table	Purpose	Updated By	Notes
`ROSTER_WARBYPARKER`	WBP XO agent roster — name, team, IS_XO_AGENT	core_ph_hr_roster_daily DAG	Source: BambooHR / PeopleHub
`ROSTER_CONDENAST`	CND XO agent roster	core_ph_hr_roster_daily DAG	Source: BambooHR / PeopleHub

GOLD Schema¶

Object	Purpose
(Shared Gold dimensions — to be populated as shared dims are built)

OPERATIONS (Legacy — Multi-Client Analytics DB)¶

This database contains legacy multi-client analytics objects. Actively maintained for ad hoc analysis and reporting; no _DEV variant exists — all queries run against prod data.

Schema	Objects	Purpose	Naming pattern
`STAGING`	~9 dynamic tables	Intermediate transformation layer	`STAGE_{CLIENT}_{OBJECT}`
`ODS`	~15 dynamic tables	Cross-client operational data store	`XOS_`, `{CLIENT}_`
`AGGREGATE`	~45 dynamic tables + ~36 views	Pre-computed metrics and dashboards	`{CLIENT}_{METRIC}`, `{CLIENT}_RPT_*`
`BAMBOOHR`	~7 dynamic tables	Employee roster source (feeds CORE_DB)	`ROSTER_{CLIENT}`

Do not create new objects in OPERATIONS. For new analytical surfaces, build in the medallion architecture (WBP_DB/CND_DB GOLD layer). Existing objects are read-only for analysis.

Airflow Connections → Snowflake Mapping¶

Airflow Connection ID	Snowflake Database	Role	Used By
`snowflake_loader_prod`	WBP_DB, CND_DB, CORE_DB	XO_DATA_LOADER	xo-foundry load tasks
`snowflake_dbt_prod`	WBP_DB, CND_DB, CORE_DB	XO_DBT_ROLE	dbt transformations (via Cosmos)
`snowflake_loader_dev`	WBP_DB_DEV, CND_DB_DEV	XO_DATA_ENGINEER_DEV	Local development

Naming Constraints¶

Before creating any new Snowflake object, verify: 1. Table name doesn't already exist in this inventory 2. Follows naming convention from ADR 009 3. Lives in the correct database — per-client data in client DB, shared data in CORE_DB 4. Schemachange migration file created — DDL must go through version control