Summary

The common.ai SQLToolset metadata tools (list_tables, get_schema) operated against a single schema, so an agent over a multi-schema warehouse (common on Snowflake) could not discover tables across schemas. With no schema set and schema-qualified tables, list_tables introspected a literal "None" schema (SHOW TABLES IN SCHEMA "DB"."None") and failed.

allowed_tables entries may now be schema-qualified ("SCHEMA.TABLE"). list_tables introspects each referenced schema and returns the matching tables fully qualified, and get_schema routes each qualified name to its own schema.

How it works

• Qualified entries are grouped by their schema; list_tables queries each schema and emits SCHEMA.TABLE names, filtered to the allow-list. get_schema("SCHEMA.TABLE") splits the identifier and passes the schema to the hook.
• Unqualified entries and the allow-all case keep the previous single-schema behaviour using the default schema, so existing setups are unchanged.
• Table-name matching is case-insensitive, because databases reflect identifiers in their own case (Snowflake reflects unquoted names lowercased) and a byte-exact match against uppercase entries would otherwise return nothing. The filter is a visibility hint, not access control, so this is safe.
• Results are de-duplicated by (schema, table) so a table reachable both qualified and via the default schema is listed once.

Usage

SQLToolset(
    db_conn_id="snowflake_hq",
    allowed_tables=["MODEL_ASTRO.DEPLOYMENT_IMAGE_DETAILS", "MODEL_CRM.SF_ASTRO_ORGS"],
)

Gotchas

• The schema name passed to the database must match its own identifier casing (uppercase on Snowflake, lowercase on Postgres); case-insensitive matching covers the allow-list filter against reflected names, not the schema lookup itself.
• Cross-database tables are out of scope: use a separate toolset whose connection points at that database.