2.0.18

What's Changed

• Add Variance analyzer with DQDL rule support and tests by @kyraman in #693
• Fix iceberg column pruning by @sudsali in #685
• Bump org.apache.spark:spark-core_2.12 from 3.5.0 to 3.5.7 by @dependabot[bot] in #675
• Add Range analyzer with DQDL rule support and tests by @kyraman in #694
• fix: Add least-privilege permissions to CI workflow by @sudsali in #696
• Add ZerosCount analyzer with DQDL rule support and tests by @kyraman in #697
• feat: Add AI-powered issue triage and PR review bot using Amazon Bedrock by @sudsali in #692
• Add Skewness analyzer with DQDL rule support and tests by @kyraman in #698
• feat: Auto-review fork PRs via pull_request_target by @sudsali in #699
• [Chore] Address bot feedback by @kyraman in #700
• feat: Post 'no issues found' comment on clean PRs instead of skipping by @sudsali in #701
• Kurtosis analyzer with DQDL rule support and tests by @kyraman in #702
• Add IQR analyzer with DQDL rule support and tests by @kyraman in #704
• Deequ Bot KB auto update by @sudsali in #705
• fix: reduce PR review false positives, increase context budget by @sudsali in #707
• Add HistogramBinned analyzer support for custom edges by @kyraman in #706
• fix: serialize Histogram where filter in MetricsRepository (#271) by @sudsali in #708
• fix: column escaping for special characters, bigint overflow, and ColumnProfiler performance by @sudsali in #711
• fix: incremental PR review, auto-approve, and bot operational improvements by @sudsali in #710
• fix: KB generator two-tier approach — all 200 files represented within budget by @sudsali in #713
• fix: Histogram column name collision and Uniqueness toString display by @sudsali in #712
• fix: remove secret_id from log to satisfy CodeQL by @sudsali in #714
• Extract null bin to separate nullCount field in DistributionBinned by @kyraman in #715
• Add overflow bins support to HistogramBinned by @kyraman in #716
• Drop out-of-range values for numeric histograms instead of mixing wit… by @kyraman in #718
• fix: two-phase PR review reduces false positives via falsification by @sudsali in #719
• Add tail support to categorical Histogram by @kyraman in #720
• Emit nullCount in HistogramBinned flatten() for metrics output by @kyraman in #721
• add support for duplicateRowCount by @awang923 in #723
• Fix NPE: encode bin edges in BinnedFrequencies state instead of mutab… by @kyraman in #722
• Add DuplicateRowCount to supported DQDL rules in README by @kyraman in #727
• Produce single bin for constant-value columns instead of zero-width duplicates by @kyraman in #726
• Update version in pom.xml to 2.0.17-spark-3.5 by @github-actions[bot] in #725
• fix: drop unused cachePoint from invoke() to avoid wasted cache-write premium by @sudsali in #728
• feat: agentic 3-stage PR review pipeline (Investigator + Critic + Reporter) behind flag by @sudsali in #724
• fix: harden agent tools against path-scope leaks and symlink escapes; raise Investigator turn floor by @sudsali in #730
• fix: persist full investigator narrative on every pipeline artifact by @sudsali in #731
• fix: agentic pipeline architectural contract — forced commit phase, aggregate text, Critic backstop by @sudsali in #732
• fix: surface commit_phase_ran in per-stage artifact metrics by @sudsali in #733
• perf: cachePoint on issue invoke() and converse_with_tools messages tail by @sudsali in #734
• perf: per-stage model_id override (Reporter on Haiku 4.5) by @sudsali in #735
• perf: per-stage model_id override for Critic by @sudsali in #736
• perf: switch Investigator/Critic baseline to Opus 4.7 by @sudsali in #737
• address review feedback for DuplicateRowCount by @awang923 in #729

New Contributors

• @dependabot[bot] made their first contribution in #675

Full Changelog: 2.0.16...2.0.18