CSV Generator — Free Online CSV Generator

How do I generate a CSV file with custom columns and delimiters?

Open AnyTool CSV Generator and pick a mode. In Schema mode you build columns, choose a fake-data field type for each (name, email, phone, date, number, UUID and more) and a row count, then generate up to 1,000 rows of clean sample data; in Manual mode you type your own header and a small editable grid. Then set the CSV format options — delimiter (comma, semicolon, tab/TSV, pipe or custom), quote character, quoting (minimal / all / none), header row on or off, CRLF vs LF line endings, and an optional UTF-8 BOM for Excel. The output is escaped per RFC 4180, with a live preview that re-parses exactly what you copy or download. Everything runs in your browser — nothing is uploaded.

Two modes: generate sample rows from a schema, or type your own header + grid
Pick the delimiter — comma, semicolon, tab (TSV), pipe or a custom string
Quote character and quote mode (minimal / all / none), header toggle, CRLF or LF, UTF-8 BOM
RFC 4180-correct quoting: fields with the delimiter, a quote or a newline are wrapped and quotes are doubled
Reuses 30+ fake-data field types; copy or download .csv / .tsv — 100% in your browser

What is

CSV (Comma-Separated Values)

CSV is a plain-text tabular format where each line is a record and fields are separated by a delimiter (classically a comma). RFC 4180 defines how to handle special characters: a field containing the delimiter, the quote character or a line break is enclosed in double quotes, and a quote inside such a field is escaped by doubling it. Records are separated by CRLF, though LF is widely accepted. A CSV file may begin with a header row naming the columns.

Generators

Related terms

DelimiterRFC 4180TSVUTF-8 BOMQuoting / escaping

Frequently Asked Questions

AnyTool CSV Generator separates the data from the formatting. For the delimiter you can pick a comma, a semicolon (common in European locales), a tab to produce a TSV, a pipe, or type any custom string. You also choose the quote character (double or single), the quote mode — minimal quotes only the fields that need it per RFC 4180, all wraps every field, none disables quoting — whether to emit a header row, the line ending (CRLF, the RFC 4180 default, or LF for Unix tooling), and whether to prepend a UTF-8 byte-order mark so Excel detects the encoding. Every option re-serializes the output instantly.

You can use a comma, semicolon, tab (TSV), pipe or a custom delimiter, choose the quote character and quote mode, toggle the header row, pick CRLF or LF line endings, and add a UTF-8 BOM.

Quoting is RFC 4180-correct and aware of the delimiter and quote character you chose. In minimal mode a field is quoted only when it actually contains the active delimiter, the quote character, a CR or an LF; the field is then wrapped in the quote character and every occurrence of that character inside it is escaped by doubling. This means values with embedded commas, semicolons, quotes or multi-line text round-trip cleanly into Excel, Google Sheets, pandas, databases and any compliant parser. A live preview re-parses the exact bytes you will copy or download so you can confirm the escaping before exporting.

It follows RFC 4180: any field containing the delimiter, the quote character, a carriage return or a line feed is wrapped in the quote character, and a quote inside the field is doubled.

Excel uses your system list separator to split CSV columns, and in many European locales that separator is a semicolon, not a comma — so a comma-delimited file lands entirely in column A. The fix is to set the delimiter to semicolon in the CSV options (or open the file via Data → From Text and pick the delimiter manually). Enabling the UTF-8 BOM option also helps Excel recognise UTF-8 so accented or non-Latin characters are not garbled. These are quirks of Excel and locale, not of the file — the output itself is standards-compliant.

Your Excel locale probably expects a semicolon delimiter rather than a comma. Switch the delimiter to semicolon, and tick the UTF-8 BOM so accented characters display correctly.

All generation and formatting happen in your browser with no server round-trip, no CDN and no upload, so your schema, your typed cells and the resulting file never leave your device and the tool works offline once cached. The schema mode reuses the same bundled clean word lists as the Fake Data Generator, so the rows are synthetic and fictional — invented names and companies, example.com / test-domain emails, 555-01xx phone numbers and clearly-marked TEST card and IBAN values — ideal for imports, fixtures and demos but never real PII. In manual mode the data is exactly what you enter.

Yes, it is 100% client-side — nothing is uploaded. Schema-mode data is synthetic and obviously fake (example.com emails, test card numbers); manual data is only what you type.

Detailed Explanation

Methodology

How the CSV Generator Builds a File

AnyTool CSV Generator deliberately separates the data from the formatting and runs both entirely in the browser. It offers two modes. Schema mode reuses the same dependency-free fake-data engine as the Fake Data Generator: you build an ordered list of named columns, give each one of 30-plus field types (names, emails, phones, addresses, companies, dates, numbers, booleans, UUIDs, colours, IPs, prices and clearly-marked test card / IBAN values), pick a row count capped at 1,000, and optionally lock a seed for reproducible rows. Manual mode instead gives you a small editable grid — type your own header and cells, including multi-line cells. Either way the resulting two-dimensional table is handed to one shared serializer that applies your CSV format options, so the same RFC 4180 logic governs sample data and hand-typed data alike.

Two modes: schema-driven sample rows, or a manual header + editable grid
Schema mode reuses the fake-data engine — 30+ field types, presets, up to 1,000 rows, optional seed
Manual mode supports multi-line cells that are escaped on export
One shared serializer formats both modes identically
All generation and formatting is client-side — no server, no CDN, no upload

Technical Details

Delimiters, Quoting and RFC 4180 Escaping

The CSV-specific controls are what set this tool apart from a plain data generator. You choose the delimiter — comma, semicolon (common in European Excel locales), tab to emit a TSV, pipe, or any custom string — the quote character (double or single), and the quote mode: minimal quotes only fields that need it, all wraps every field, none disables quoting. You can toggle the header row, switch line endings between CRLF (the RFC 4180 default) and LF (Unix), and prepend a UTF-8 byte-order mark so Excel detects the encoding. Escaping is RFC 4180-correct and delimiter-aware: any field containing the active delimiter, the quote character, a carriage return or a line feed is wrapped in the quote character, and each quote inside it is doubled. A live preview re-parses the exact output so you can confirm the escaping before copying or downloading a .csv or .tsv file.

Delimiter: comma, semicolon, tab (TSV), pipe or a custom string
Quote character (double/single) and quote mode: minimal / all / none
Header toggle, CRLF vs LF line endings, optional UTF-8 BOM for Excel
RFC 4180 escaping: delimiter/quote/newline fields are quoted; quotes doubled
Live re-parsed preview; copy or download .csv / .tsv

Privacy & Security

Privacy, Excel Locales and the Synthetic-Data Note

Because everything happens in the browser, neither the schema you build, the cells you type nor the resulting file are ever uploaded, logged or stored — closing the tab discards them and the tool works offline once cached. The page is honest about two practical caveats. First, Excel quirks: if a comma-delimited file opens with all data in one column, the local list separator is probably a semicolon, so switching the delimiter (and enabling the UTF-8 BOM for accented characters) fixes it — a property of Excel and locale, not of the standards-compliant file. Second, schema-mode rows are synthetic and obviously fake: invented names and companies, example.com / test-domain emails, 555-01xx fictional phone numbers, and clearly-marked TEST card and IBAN values, intended for imports, fixtures and demos but never as real PII. Manual-mode data is only ever what the user types.

Schema, typed cells and file never leave the device; works offline once cached
Excel may need a semicolon delimiter and a UTF-8 BOM in some locales
Schema rows are synthetic (example.com emails, TEST cards) — not real PII
Manual data is exactly what the user enters

CSV generation: in-browser (AnyTool) vs typical online CSV tools
Capability	AnyTool	Typical online CSV tools
Processing	Runs entirely in your browser	Often server-side / API-gated
Delimiter control	Comma, semicolon, tab, pipe, custom	Usually comma only, or limited
Quoting	Minimal / all / none, RFC 4180-correct	Often fixed, not always compliant
Line endings & BOM	CRLF or LF, optional UTF-8 BOM	Rarely configurable
Data source	Schema sample data or your own grid	One or the other, seldom both
Field types	Reuses 30+ fake-data types	Varies; often a small fixed set
Privacy	Never uploaded, works offline	May transmit or log your data
Cost / signup	Free, no signup	Often gated or rate-limited

AnyTool builds and formats the CSV locally and uploads nothing.

Cite This Page

Title: CSV Generator (schema- or grid-driven CSV with configurable delimiter, quoting, line endings and UTF-8 BOM, RFC 4180-correct, 100% client-side)
Author: AnyTool
Published: January 1, 2026
Modified: February 8, 2026
Section: Generators
URL: https://www.anytool.tech/csv-generator

Formatted Citation

AnyTool. "CSV Generator (schema- or grid-driven CSV with configurable delimiter, quoting, line endings and UTF-8 BOM, RFC 4180-correct, 100% client-side)." AnyTool, January 1, 2026. https://www.anytool.tech/csv-generator

A privacy-first, 100% client-side CSV generator that separates the data from the formatting and runs both in the browser with no server round-trip, no CDN and no upload, so the schema, the typed cells and the resulting file never leave the device and the tool works offline once cached. It offers two modes sharing a single serializer: a schema mode that reuses the same dependency-free fake-data engine as the Fake Data Generator (an ordered list of named columns, each assigned one of more than thirty field types across personal, contact, address, business, text, numeric, date-and-time and technical groups plus clearly-marked test card and format-only IBAN values, one-click User/Product/Address-book/Employee/Transaction presets, a row count capped at 1,000 and an optional seed driving a small mulberry32 generator for reproducible rows), and a manual mode offering a small editable grid where the user types their own header and cells including multi-line cells. The differentiator is CSV-specific control: the delimiter can be a comma, a semicolon (common in European Excel locales), a tab to emit a TSV, a pipe or any custom string; the quote character can be double or single; the quote mode can be minimal (quote only fields that need it), all (wrap every field) or none; the header row can be toggled; line endings can be CRLF (the RFC 4180 default) or LF; and an optional UTF-8 byte-order mark can be prepended so Excel detects the encoding. Escaping is RFC 4180-correct and delimiter-aware — any field containing the active delimiter, the quote character, a carriage return or a line feed is wrapped in the quote character and each embedded quote is doubled — and a live preview re-parses the exact output so the escaping can be confirmed before the result is copied or downloaded as a .csv or .tsv file. The page is explicit about two honest caveats: Excel may place a comma-delimited file entirely in one column when the local list separator is a semicolon (fixed by switching the delimiter and enabling the UTF-8 BOM for accented characters, a quirk of Excel and locale rather than of the standards-compliant file), and schema-mode rows are synthetic and obviously fictional (invented names and companies, example.com / test-domain emails, 555-01xx phone numbers, a Luhn-valid but non-real 4111-style test card and a TEST-marked IBAN) intended for imports, fixtures and demos but never as real PII, while manual-mode data is only ever what the user types.