Merge CSV files and review duplicates.
Choose files, confirm their columns and keys, select what to keep, then take away cleaned and excluded-row CSV files. Everything stays in this browser.
Choose CSV files
UTF-8 CSV only, up to 20,000 rows and 100 columns per file. Source files are never changed.
Approved-value dictionary (optional)
Only exact cell values are replaced by column. The dictionary stays in this page's memory.
No dictionary loaded.
Selected files
Choose CSV files.
Duplicate review (first 40 rows)
| File | Source row | Key | Group | Decision |
|---|
Take away the files
Check that input rows = cleaned rows + excluded rows before saving.
Compare a recipe change (optional)
Apply the old and new recipes to the same input CSV and review changed rows and newly rejected rows. The new rule is applied and saved only after approval.
Choose the input CSV and old/new recipes.
Change list (first 40 rows)
| Source row | Old | New | Category | Changed fields |
|---|
Apply repeat-delivery recipes to multiple CSVs (optional)
Load a recipe-set JSON, assign one recipe explicitly to every file, then save destination-scoped CSVs and a manifest in one ZIP. Filenames are never used for matching.
Up to 100,000 data rows. Files stay in this browser.
Choose a recipe-set JSON.
Resubmit rejected rows only (optional)
Choose the original CSV and the destination result CSV, then review and edit failed rows only. Successful rows are not exported; unknown rows stay separate.
Choose both CSV files.
Review and edit failed rows
Shown separately as unknown
Relation-preserving pseudonymized copy (optional)
Keep customer/order relations while explicitly choosing keep, pseudonymize, fixed value, or remove for every column. Pseudonyms are not a legal anonymization or external-sharing guarantee.
Choose a schema JSON and CSV files.
Pseudonymization preview
Columns marked keep retain original values. Pseudonymization does not guarantee anonymity, legal compliance, or safe external-AI sharing.
Repair broken references and deliver related tables (optional)
Declare parent tables, child tables and keys in schema JSON, then assign each CSV explicitly. Orphans, duplicate primary keys, empty keys and cycles stop safely; save the corrected table set as a ZIP.
Load a schema JSON and CSV files.
Review and repair issues
Convert dates, decimals, or units explicitly (optional)
Choose the input and delivery formats, review each converted value, then save. Ambiguous, invalid, and unsupported values are never guessed.
Grouping is checked exactly: choose ASCII space (U+0020) or nonbreaking space (U+00A0) to match the source. Invalid values stay unchanged and block saving.
Choose one CSV file.
Review converted values
Build delivery JSON from parent and child CSVs (optional)
Assign a recipient schema or saved mapping manifest and CSVs explicitly, review parent/child references, then save nested JSON and a reusable mapping manifest. Nothing is sent externally.
Load a schema JSON and CSV files.
Review nested payload
Review Shopify product update CSV safely (optional)
Compare an existing export and a proposed update in the browser, hold blank values instead of erasing data, and save only explicitly adopted changes. Nothing is sent to Shopify.
Choose an existing export and a proposed update CSV.
Review Shopify update candidates
Blocking issues
Warnings
Rows to adopt
Reconcile invoices and payments (optional)
Choose invoice and payment columns explicitly. Review split payments, one payment across multiple invoices, and unmatched rows; save only an explicitly confirmed local report. Nothing is sent to a financial institution.
Choose an invoice CSV and a payment CSV.
Review allocations
Blocking issues
Notes and suggestions
For each payment, add invoice and amount rows. An exact reference is only a suggestion; it is never adopted automatically.
Review similar duplicate candidates (optional)
Choose a comparison column and decide each candidate as same, different, or hold before saving a cleaned CSV. Nothing is merged automatically.
Choose one CSV file.
Decide candidates
Reproduce CSV delivery errors with synthetic data (optional)
Create fixed synthetic cases and reproduction steps locally so you can discuss a delivery error without sharing source data. Select cases, then explicitly generate and save.
Select one or more cases.
Input, filenames and results stay in temporary browser memory until an explicit save. No automatic deletion or entity matching is performed.
Turn team time into a client report
Filter member work CSVs by project and period, choose the columns to share, review the rows, then save a report CSV. Everything stays in this browser.
Header: member,project,start,end,minutes,description. Timestamps must include a timezone.
Choose CSV files.
Review before sharing
| Member | Project | Date | Minutes | Description | Source file | Decision |
|---|
Partition one CSV by recipient (optional)
Choose the recipient key column, recipients, and columns explicitly, then create one CSV per recipient. Check source, distributed, and excluded row counts before saving. No guessing, deduplication, or external sending.
Choose partition rules
Outputs before sending
Supported scope
- Reordered columns are mapped to the first file's headers. Different headers stop safely.
- Compare one to three key columns and choose first, last or all duplicate rows to keep.
- Original cell text is preserved, with cleaned and excluded-row CSVs kept separate.
This does not perform customer matching or guarantee destination acceptance. Review the output and destination requirements.