Make the next
handoff clearer.
Understand the CSV in front of you.
Explain its structure before it travels.
Read in this tab. No upload, account or saved browser copy. Refreshing clears your work.
Start with a file
UTF-8 · up to 2 MiBOr drop one file here. It is read locally.
Download a sample CSV to try the file workflow.
Files retain their original line endings until edited. Pasted or edited text uses browser-normalized LF newlines. Changes need another inspection.
Choose an example, a file or paste text to begin.
Every handoff starts
with a shared understanding.
Inspect a file to see the record shape, compare parsing choices and locate problems.
Inspection
Compare delimiters
No automatic choice. A one-column result can be valid or use the wrong delimiter.
Findings
Source context
Original text around this physical source line. Included here for inspection; omitted from reports.
Record preview
Cell contents render as text. Quotes and multiline fields are parsed; values are not evaluated.
Hand over the structure.
Download parsing choices, counts and findings. Cell values, header names, source excerpts and the filename are omitted.
Counts and source locations remain. This is a structural check, not a data-quality or spreadsheet-safety certificate.
A clear read, not a silent repair.
Choose the interpretation
CSV dialects vary. Pick a delimiter, decide whether there is a header and compare the results. Quoted newlines and doubled quotes are supported.
Keep the original intact
RowBrief checks structure only. It does not correct data, infer types or execute formulas. Fatal quoting or size errors stop inspection; partial results are labeled.
Share a smaller report
The report omits content and names. Review it before sharing. No files, reports or preferences are stored by the app.
Limits and practical details
Strict UTF-8 only, with or without BOM. Up to 2 MiB, 10,000 parsed records and 100 fields per record; only the first 200 findings are listed. Blank records still count toward the record limit. Source lines and columns are one-based; columns count Unicode code points. CRLF is one line break. A leading BOM is excluded from visible column positions.
Headers are checked for empty values and exact duplicates. Field counts are compared with the first included record even when the header option is off. No business-rule, data-type, encoding-conversion, formula-safety or universal spreadsheet-compatibility check is provided. A first page load requires a connection unless you serve your own local copy. No offline cache is installed.