This pull request removes the sample dataset selection logic from the frontend onboarding flow and centralizes sample data ingestion control to a backend environment variable. The onboarding UI no longer presents an option to ingest sample data, and the ingestion behavior is now controlled via the INGEST_SAMPLE_DATA flag in the backend configuration. Additionally, error handling for sample data ingestion failures has been improved to roll back onboarding steps more reliably.

Configuration and feature flag changes:

• Added the INGEST_SAMPLE_DATA environment variable to .env.example, docker-compose.yml, and Helm charts, allowing ingestion of sample data to be toggled at startup via backend configuration. [1] [2] [3] [4] [5]

Frontend onboarding flow simplification:

• Removed all references to the sample dataset selection (sampleDataset, setSampleDataset, and related handlers) from onboarding components including AdvancedOnboarding, AnthropicOnboarding, IBMOnboarding, OllamaOnboarding, and OpenAIOnboarding. [1] [2] [3] [4] [5] [6] [7] [8] [9] [10] [11] [12] [13] [14] [15] [16] [17] [18]

Error handling improvements:

• Improved task failure detection and onboarding rollback logic in onboarding-card.tsx to handle both task-level and file-level failures, and to trigger rollback mutation instead of manually stepping back. [1] [2] [3]

Code cleanup:

• Removed unused imports and state related to sample dataset selection from onboarding-card.tsx. [1] [2] [3]

These changes streamline the onboarding experience and ensure sample data ingestion is managed consistently via backend configuration.