Skip to content

Free · Edit locally · No signup

Edit CSV files online

Edit CSV in a virtualized browser workspace with local paged storage. Inspect, sort, filter and download a copy; limits depend on available storage.

Open and edit CSV

Start with your CSV

Open your file in the CSV editor with this tool already selected. Preview the result before downloading.

  1. Choose one CSV or TSV file.
  2. Edit cells, filter or sort your table.
  3. Download the complete edited file.

Free · No signup · File contents stay in your browser. Open files up to 1 GiB with browser storage.

Monthly product sales

8 records · 6 columns

sample-sales.csv · Preview: first 4 of 8 records

Monthly product sales CSV sample data
monthregionproductunitsrevenuereturned
2026-01-01NorthField Notes1423408.00false
2026-01-01SouthCanvas Tote962688.00false
2026-02-01EastDesk Tray, Small742812.00true
2026-02-01WestField Notes1653960.00false

Illustrative fictional data. Open the sample to try the tool.

From file to finished copy

Edit your CSV without losing its text values

Choose a local CSV or TSV file, paste text, or open the sample. The shared CSV workspace keeps your table available while you inspect long cells, change values, filter rows and sort columns.

Use Download copy to download the complete edited table. Filters and sorting change the view; they do not remove records from the saved file. When an imported document is unchanged, including after Undo restores its original content state, the copy retains its original bytes. For strict UTF-8 originals up to 8 MiB, 100,000 data records and 1 million cells, cell-only changes retain untouched field spelling, BOM and record endings when you change only cell values without changing headers or adding, removing or rearranging rows and columns. Other edited copies preserve literal cell values but can normalize quoting and line endings; download feedback identifies the formatting used.

Selected files have an upper limit of 1 GiB, 5 million data records and 2,000 columns. Available browser storage can impose a lower limit. Very wide records, oversized fields and the 100,000 stored-page cap can also stop import before those limits. Decoded fields are limited to 8 MiB and decoded records to 16 MiB; stored and displayed row pages must also fit their 16 MiB encoded-page limit. Pasted CSV text is limited to 50 MiB. For larger input, choose a local file instead.

The editor parses in a worker and stores row pages in browser storage while showing only the visible rows and columns. Browser storage is temporary working space, not a backup: download a copy before closing the editor.

Prefer an installed app? Get CSV Editor for Windows, Mac or Linux. Check the current store listing for that platform's released capabilities and limits.

Use Clean CSV to trim fields and remove matching records, Validate CSV to check syntax and column rules, or CSV to Excel for a spreadsheet download.

For a known input before editing your own file, inspect the documented CSV samples in the CSV viewer.

In Tools → Recipes, build repeatable column steps. Optional assistance, when available, can draft settings from your goal and explicitly included column metadata. Review the draft, preview the complete result, then separately Apply or export. See how optional assistance processes context.

In Tools → Views, choose Wrap text in taller rows to read several lines in each cell. Row height stays bounded for large tables; Inspector shows and edits the complete value. Switching the display does not change file contents.

Dedicated tools can clean, map, split, validate or convert your data. Each tool previews its result before you apply changes or download a separate copy. Task calculations and converters retain their own processing limits, even when the editor can open a larger file.

Plain answers

Edit CSV questions

What can I do with this CSV editor?

Edit your file locally and save a new copy.

Does my CSV leave my device?

File selection, editing and these task operations run in your browser. Optional recipe assistance, when available, sends your goal and only the column metadata or sample values you explicitly include to its configured provider. Your file is not attached automatically; see Privacy for processing details. Manual editing requires no account or file upload. Download a new copy to keep your result; refreshing restores the original input, not unsaved edits.

How do I export filtered CSV rows to Excel?

Open your CSV in the browser editor, set the search, column filters and sort order, then open Inspector. Under Current view, choose Export current view as XLSX. On a narrow screen, open Show Tools first. The separate workbook contains every matching row in that order, with the original headers; a cell selection does not limit this export. For example, filtering an ID column to 001 keeps the identifier 001 as text. Formula-looking values such as =SUM(A1) also remain text cells rather than executable formulas. The open CSV, its save state and Undo history are unchanged. This bounded export supports 100,000 data rows, 2,000 columns, 1 million cells including headers, 32,767 characters per cell and 64 MiB of text or output. Unsupported control characters and malformed Unicode are refused. Cancel during preparation discards the result; workbook creation cannot be interrupted at every instruction. Filter to a smaller view if a limit is reached. These steps describe this browser workspace; installed apps and extensions need their own tested release support.

Can AI help me build a CSV recipe?

When assistance is available, open Recipes in the editor Tools. Capture selected column metadata, review and explicitly include it, then describe your goal. Review the draft settings before importing them. Complete-source Preview and a separate Apply or export are still required; unsupported drafts are refused. You can build recipes manually without assistance.

Can I save and reopen my CSV workspace as a project?

In the browser editor on a wide screen, choose Project beside the document. On a narrow screen, choose Show Tools, then Project in Tools workspace. Enter a Project name and choose New project to capture the current document view. Save project copy downloads a .loam-project.json file containing document references, column names, filters, sorting and pane layout. Save your edited data separately: the project contains no data cells and cannot restore unsaved cell edits or grant access to source files. To return later, choose Open project and select the saved project file. Open a saved document, or use Choose source for that document to select its source again. Review source binding before choosing Open with reviewed columns. If the source or unsaved state differs, recover your edited data separately or deliberately choose Open source and reset saved view. Cancel source binding keeps the current document. Projects support up to 64 document references and 2 MiB of metadata. These instructions describe the browser workspace; installed apps and extensions need their own host support.

Can I export only selected CSV cells?

Yes. Select a rectangular range, open Inspector and choose CSV, TSV or XLSX under Selection export format, then Export. On a narrow screen, open Show Tools first. For separate rows, use row-number buttons with Ctrl/Cmd to toggle individual rows or Shift to extend. For columns, use their actions menus to select or toggle columns. Selected rows include all columns; selected columns include every row in the current filtered/sorted view. For example, selecting rows 1 and 3 includes those two rows and leaves row 2 out. All formats include exact raw headers, including duplicate or blank names. CSV and TSV are separate UTF-8 copies supporting 100,000 selected rows, 2 million cells and 64 MiB of encoded output. TSV uses tabs and quotes fields containing tabs, line breaks or quotes. Spreadsheet applications may evaluate formula-like text in CSV, TSV or clipboard copies. XLSX stores every value as text, including identifiers such as 001 and formula-like strings such as =SUM(A1). It supports 100,000 selected rows, 2,000 columns, 1 million cells including headers, 32,767 characters per cell and 64 MiB of text or output. Unsupported controls or malformed Unicode refuse the complete XLSX. Clipboard copies are limited to 1 MiB. Cancel during preparation or select a smaller result if a limit is reached; no partial copy is offered. Workbook creation cannot be interrupted at every instruction. Changing the source, selection, view or format invalidates preparation. Export leaves source history and save state unchanged. These instructions describe this browser workspace; installed apps, extensions and the TSV product need their own tested release support.

Can an unchanged download preserve the original CSV bytes?

Yes. For an imported file, Download copy uses the original admitted bytes while the document is still at that exact content state, including after Undo restores it. Search, filters and view sorting do not change the source copy. This preserves original quoting, BOM, encoding, mixed line endings and final-newline choice. For strict UTF-8 originals up to 8 MiB, 100,000 data records and 1 million cells, cell-only edits preserve unchanged field spelling, BOM, mixed record endings and final-newline choice when you change only cell values without changing headers or adding, removing or rearranging rows and columns. A changed cell uses safe CSV quoting; adding a value in a formerly missing ragged field extends only that record as needed. Structural edits, unsupported grammar, other encodings and larger sources use normalized output, which can normalize quoting and line endings. Download feedback identifies preserved or normalized formatting. Edited lexical output is limited to 50 MiB. The browser downloads a separate copy and never overwrites your original file.

Can I compare or join every source row?

In Tools, open Analysis. Under Understand every source row, choose Compare every source row or Join every source row in Analysis type. Choose another UTF-8 CSV or TSV as Comparison file, then pair one or more key fields in order. Every key field must be nonblank and the combined key unique in each file. For TSV, choose literal quotes or spreadsheet-style quoted fields to match the input. Comparison requires unique, exact column names in both files; their order may differ. Preview complete comparison reports change kind, key and source/imported data-row positions; it does not list before-and-after cell values. A join can keep every source row or only matching rows; unmatched imported values are blank. Joined output column names must also be unique. Both operations include rows hidden by filters and sorting. Export every result row creates a separate CSV without applying edits or changing either source or Undo history. The complete result must fit 1 million combined keys, 64 MiB of retained state, 250,000 result cells and 16 MiB of output; exceeding a budget refuses the result rather than returning a partial copy. The separate Compare CSV task and loaded-table tools retain their own smaller limits. These instructions describe the browser workspace; installed apps and extensions need their own host support.

Do the editor and task have the same file limits?

The editor opens files up to 1 GiB and 5 million records, subject to browser storage and decoded-record limits. Focused calculations and converters have their own processing limits, which are shown by each tool.

Other ways to work with your CSV

Use these local examples when you prefer a script or SQL. They create separate results and explain where their behavior differs from this browser tool.

Update a CSV value with Python

Writes a separate output.csv, changing the label only for records whose id is exactly 001.

Before you start

  • Python 3.10 or later. Run in a working folder with a backup of your source; output files may be replaced unless opened in exclusive mode.
  • Place input.csv in that folder. Examples use UTF-8, a comma delimiter and a header row. Column-specific recipes use id,label,amount,status,note; adapt names to your file.
Run the complete Python example
python3 - <<'PY'
import csv

with open("input.csv", encoding="utf-8-sig", newline="") as source:
    table = list(csv.reader(source, strict=True))
if not table:
    raise ValueError("Expected a header row")
headers, rows = table[0], table[1:]
if any(len(row) != len(headers) for row in rows):
    raise ValueError("Every record must match the header width")

id_column = headers.index("id")
label_column = headers.index("label")
for row in rows:
    if row[id_column] == "001":
        row[label_column] = "Updated label"

with open("output.csv", "w", encoding="utf-8", newline="") as target:
    writer = csv.writer(target)
    writer.writerow(headers)
    writer.writerows(rows)
print(f"Wrote {len(rows)} records to output.csv")
PY

Limits and output

  • CSV has no stored type schema. Python keeps fields as strings until an example explicitly converts a value. These recipes do not add spreadsheet formula protection; import sensitive-looking strings as text in spreadsheets.
  • This recipe retains the parsed table in memory; use a streaming approach for files that do not fit.

Documentation: Python CSV reader and writer

Use DuckDB for CSV editor

Run SQL locally with the DuckDB command line. Read fields as text, then cast only the columns used in calculations.

Before you start

  • Install the DuckDB command-line client using its official installation guide.
  • Run beside input.csv. Named-column examples expect id,label,amount,status,note.
Run DuckDB SQL
duckdb <<'SQL'
CREATE VIEW source AS SELECT * FROM read_csv('input.csv', header = true, all_varchar = true, force_not_null = ['id', 'label', 'amount', 'status', 'note']);
COPY (SELECT id, CASE WHEN id = '001' THEN 'Updated label' ELSE label END AS label, amount, status, note FROM source) TO 'duckdb-output.csv' (HEADER, DELIMITER ',');
SQL

Limits and output

  • Adapt column names and casts to your input. DECIMAL(18,4) rounds to four fractional digits; values outside its range become NULL with TRY_CAST. SQL output order is not promised without ORDER BY; equal sort keys need an explicit tie-breaker for a deterministic order.
  • These recipes write ordinary CSV without spreadsheet formula protection.

Documentation: DuckDB CSV options and installation; DuckDB COPY export