- Max articles per query now counts new articles after duplicates are removed. Before, an article already saved by an earlier query or day used up part of the limit, so runs returned fewer rows than asked for.
- Articles are saved in batches instead of one by one, and 10 links are resolved at a time instead of 5. A 3-day range with resolved URLs went from 65 to 35 seconds.
- Queries or topics with no results are named in the final status message.
- Repeated queries run once. Topic names are accepted in any case, and an unknown topic stops the run with a clear message instead of being skipped.
- A blank language or country falls back to en / US.
First release: keyword and topic scraping in any edition, real publisher URLs, day-by-day date ranges past the 100-result cap, related coverage, deduplication.
