Transformation output fails when writing to dev-branch-only table: bucket not assigned to development branch

[jakubsebek-kosik]

Summary

When running a Snowflake transformation on a dev branch, the job fails if the output table only exists in the dev branch (i.e. it has no production counterpart). The transformation runner attempts to CREATE the table and the Keboola Storage API returns an error that the bucket is not assigned to any development branch.

Error Message

Trying to create a table in the development bucket "in.c-example-bucket" on branch "dev-branch" (ID "XXXXX"), but the bucket is not assigned to any development branch.

Steps to Reproduce

1. Create a new typed table in a dev branch that does not exist in production — e.g. in.c-example-bucket.MY_TABLE_typed (dev-branch-only).
2. Create a Snowflake transformation config on the same dev branch with an output mapping pointing to that table:

   output:
     tables:                                                                                                                                                                                   
     - source: MY_TABLE_casted                                                                                                                                                               
       destination: in.c-example-bucket.MY_TABLE_typed

3. Push the config and run the transformation job.

Expected Behavior

The transformation should write its output to the dev-branch-only typed table in.c-example-bucket.MY_TABLE_typed.

Actual Behavior

The job fails immediately during the output phase with:

Trying to create a table in the development bucket "in.c-example-bucket" on branch "dev-branch" (ID "XXXXX"), but the bucket is not assigned to any development branch.                      

Context / Investigation

• The dev-branch Snowflake schema for this bucket is "KEBOOLA_PROJECT_ID"."BRANCH_ID_in.c-example-bucket" — the bucket exists in Snowflake with a branch-ID prefix.
• The Storage API bucket ID used in the output mapping is in.c-example-bucket (no prefix), which is the same ID as the production bucket — but the table only has a branch-level copy.
• The production bucket in.c-example-bucket exists and is accessible. Only the specific table (MY_TABLE_typed) was created solely in the dev branch with no production counterpart.
• The error does not occur when the output table exists in production (even if it also has a dev-branch copy) — the transformation runner can write to those just fine.
• Removing column_data_types from the output mapping did not resolve the issue (separate problem — column_data_types is not a recognized key in transformation runner v1.5.0).

Workaround

• Download/upload approach (bypassing transformation output entirely) — not preferred.

Environment

• kbagent version: 0.25.0
• Stack: EU Central 1
• Transformation runner version: 1.5.0
• Component: keboola.snowflake-transformation