Changes in version 0.2.2 (2025-10-06)                  

  - CRAN resubmission:
      - Replaced arXiv URL with arXiv DOI
        (doi:10.48550/arXiv.2108.07258).
      - Ensured software/API names are in single quotes (e.g., 'SQL',
        'LLM').
      - Expanded acronym in Title (“LLM = Large Language Model”).
  - Documentation tidy-ups:
      - Added \value{} to generate_fake_from_schema() and clarified
        return classes.
      - Minor wording/typo fixes across vignettes and Rd files.
  - CI/docs:
      - Fixed pkgdown workflow and GitHub Pages settings.

                        Changes in version 0.2.1                        

  - CRAN fixes: expanded acronym, quoted software names, added
    references.
  - Replaced \dontrun{} with \donttest{} in examples where feasible.
  - Added \value{} sections; RNG avoids .GlobalEnv writes (uses
    withr::with_seed()).

                        Changes in version 0.2.0                        

  - Initial public submission:
      - generate_fake_data() with category/numeric/column modes.
      - generate_fake_with_privacy() (sensitive detection + strategies).
      - export_fake() (CSV/RDS/Parquet), llm_bundle() &
        llm_bundle_from_db().
      - prepare_input_data() normalization helpers; POSIXct/date
        handling.
      - Vignettes: getting started, DB workflow, privacy & validation.