Support for Granular Field Selection (Sub-field Filtering) within Nested Objects

[Lucas Leadbetter (lleadbet)]

Problem:
Currently, many Airbyte sources (such as source-github) deliver streams where data is structured as nested objects (e.g., the user field).

The current selection UI and underlying sync logic are "all or nothing" at the top-level field. There is no mechanism to select specific properties within a nested object or to flatten the object during the discovery/configuration phase to allow for sub-field selection.

Impact:

1. Security & Compliance: If a nested object contains sensitive PII (e.g., user.email) alongside necessary metadata (e.g., user.id), users are forced to either sync sensitive data or abandon the entire object to remain compliant.
2. Sync Efficiency: Syncing large, unnecessary nested objects increases storage costs and sync duration (bandwidth), especially when only a single sub-key is required for downstream analysis.
3. Usability: Users currently have no workaround to exclude these fields at the source, forcing them to handle complex cleanup and masking logic at the destination/transformation layer.

The Ask:
Enable the ability to select specific fields within an object during the connection configuration. The schema selection tree should allow users to "drill down" into objects to toggle specific properties.

Example:
In a GitHub stream, a user should be able to toggle user.id to enabled while keeping user.login or user.email disabled, ensuring only the required sub-fields are replicated to the destination.