No implicit coercion
Leading zeros, long identifiers, date-like strings and formula-like text remain visible as written.
A CSV editor that shows its work
View the complete table, edit cells and structure, filter and sort, profile columns, compare by key and build repeatable transformations without letting identifiers turn into numbers or dates.
Private early access · macOS first
DataTransformCompareIssues 6RecipeRun report
customer_iditemquantityunit_priceline_total
1001234Shipping box124.5054.00
2001235Label roll812.0096.00
3001236Paper tape243.2076.80
4001237Packing slip49.2537.00
5001238Return label67.5045.00
Text values remain textUTF-8 · comma · CRLF2 unsaved edits
Text stays text until you decide otherwise
CSVProof opens values as CSV fields, not spreadsheet guesses. Supported cleanup and normalization operations are explicit steps with a before-and-after preview.
Visible parsing
UTF-8 and BOM status, delimiter, quoting, header and newline interpretation stay visible; selected parser rules are saved with workflows.
Safer output
Consequential normalization is counted and previewed. Output is verified after an atomic write to a new file by default.
Edit in the context of the whole run
The grid is the main workspace. Comparison, transformations and issues stay connected to the rows they affect.
JobWhat CSVProof makes explicitResult
ViewRaw field and record evidence beside the parsed table; search, sort and filter without changing the file.Understand first
EditCell, row and bounded bulk changes with dependable undo and no automatic value conversion.Controlled change
ProfileEmpty values, distinct values, exact text ranges, lengths and frequency before rules are applied.Know the file
ValidateRequired values, allowed sets, patterns, uniqueness and composite keys.Explicit contract
TransformRename or remove columns; trim, replace, map, filter and normalize exact text as visible steps.Reviewable recipe
CompareOne or more record keys, duplicate/empty key checks and per-field differences.Real record identity
ReleaseReviewed exceptions, row counts, contract versions and input/output hashes.Evidence report
Transform the table without losing sight of the rows
Each supported cleanup operation becomes a named step beside a local impact preview. Rename or remove columns, replace or map exact values, normalize text, or keep and remove matching records before writing output.
The recipe records the order and exact parameters, then reports aggregate effects without copying cell values into persisted evidence.
DataTransformCompareIssues 6RecipeRun report
Trim whitespacename, item228 cells
Calculate line_total
quantity * unit_price12,481 cellsMap country codescountry_lookup.csv4 misses
Keep active recordsstatus = active93 removed
customer_idquantityunit_priceline_total
001234124.5054.00
001235812.0096.00
001236243.2076.80
00123749.2537.00
One workflow, three durable artifacts
The table is where you work. The contract, recipe and run report make that work repeatable and explainable.
Contract
How the file is parsed, which columns and keys are expected, and what should stop the run.
Recipe
The ordered transformations, assertions and output settings you approved.
Run report
What came in, what changed, which checks passed and what was written.
The next file should not restart the process
Save an approved workflow as a human-readable recipe. When the next export arrives, CSVProof checks the parser contract and reruns only after the input is understood.
- Explicit parser and output dialect
- Required fields, allowed values and unique keys
- Deterministic text, mapping, row-filter and column-structure steps
- Assertions that stop unsafe output
- Versioned run report
DataTransformCompareIssues 6RecipeRun report
Require unique customer_idcontractRequired
Trim name and addresstransform228 cells
Calculate line_totalquantity * unit_priceFormula
Join account lookupone to one match4 misses
Write UTF-8, comma, CRLFoutput formatOutput
Expected result
Rows read12,481
Rows ready12,475
Issues6
Changed cells228
Your CSV files stay on your computer
CSVProof is a native Mac app. Source files are processed locally; no account or source-data cloud is required for the core workflow. Windows and Linux may follow after the Mac release.
A focused data workbench, not a general spreadsheet.
Cleanup operations are explicit transformation steps. CSVProof leaves out charts, cell styling, cloud sync and AI guessing so the data workflow stays inspectable.
Bring repeatability to the CSV work you already do
CSVProof is currently in private early access for macOS.
Get early access