Veridelta compares two datasets on their primary keys and reports every row that differs under the rules you declare. Nothing is forgiven unless a rule says so, and the exit code tells CI whether the datasets match.

Watch the three-minute demo: Sam rewrites a nightly export of 40 accounts, Veridelta explains the noise with evidence, the one real defect fails his pull request, and the fixed export passes.

pip install veridelta
veridelta run legacy.csv modern.csv --key id

Veridelta is in alpha. Its tests run on generated data, with drift seeded on purpose, and nobody but its maintainer is known to have run it yet. Status says more. Issues and discussions are welcome.

Docs · PyPI · Discussions