CSV Sidekick

A fast, dependable CSV workspace for analysts and data scientists in Visual Studio Code.

Open CSV, TSV, TAB, and PSV files as responsive tables, inspect real-world data without breaking on messy values, and move clean selections straight into Excel, notebooks, documentation, or a downstream workflow.

Built for day-to-day data work

• Explore large extracts comfortably. Virtualized rows and columns keep wide and large files responsive, while frozen headers and Excel-style pane freezing preserve the context you need as you scan.
• Trust what you are seeing. Correct parsing handles quoted values, escaped quotes, embedded delimiters, and multiline fields. CSV Sidekick detects comma, tab, semicolon, and pipe delimiters automatically.
• Shrug off sloppy exports. When every line ends with a delimiter—a common habit of report generators—the table ignores the empty column it creates instead of showing a phantom field, and the file summary tells you it did. A column is only ignored when every record leaves it blank.
• Profile every column at a glance. Open Data profile for full-dataset column types, valid and missing-value counts, unique values, distributions, and—when values are numeric—mean, minimum, and maximum. It is a quick first pass for finding quality issues and understanding a new extract.
• Catch data-quality problems before they travel. Open Data quality to scan every record for inconsistent column counts, unclosed quoted fields, missing or duplicate ID-like values, and invalid dates. The report explains its automatic ID/date inference, and clicking an issue jumps to its source cell without changing the file.
• Open data from the real world. Automatic encoding detection supports UTF-8, UTF-16 BOMs, and a Windows-1252 fallback—useful for files exported from spreadsheets, BI tools, and legacy systems.
• Follow every field across a wide table. Each column is colorized with a repeating, color-blind-friendly 16-color palette, making records easier to scan horizontally. Numeric values are also right-aligned with tabular spacing for fast comparison.
• Sort without rewriting the file. Click a column header to cycle through ascending, descending, and original file order. Sorting understands clean numeric values—including integers larger than JavaScript can represent exactly—keeps equal values in their original order, and carries through to copied or exported data.
• Filter the complete dataset, not just loaded rows. Press Filter or Ctrl/Cmd+Shift+F to open a stack of conditions—each with its own column, comparison, and value, joined by "and"—then add or remove rows with + Condition. Right-click a cell for Filter by this value, or click a bar in Data profile, to add a matching condition in one step. Filtering runs in the parser worker, so the grid stays virtualized and sorting, profiles, quality checks, copying, Tail 10, and exports all use the same matching records.
• Find a value without hiding anything. Press Find or Ctrl/Cmd+F to search every cell in the file. Matches highlight in the grid; Enter, Shift+Enter, and F3 step through them and scroll each into view. Narrow the search to one column, match case, or match the whole cell.
• Trace one value back to its labels. Turn on Focus cell and click a cell: its whole row and column light up, along with the row number and column header at the edges of the grid, so you never lose your place in a wide extract. Column colors step aside while focus mode is on and come back when you switch it off.
• Focus on what changed. Switch to Tail 10 to inspect the newest ten records while keeping the header visible—ideal for logs and incremental exports.
• Clean data safely. Preview trimmed values and normalized headers without changing the source file; when you do need to edit, use the text-editor commands with normal VS Code Undo support.
• Select like a spreadsheet. Drag across cells to select a rectangle, Shift-click to extend it, or click the corner box above the row numbers to select every displayed cell. Click a cell and the arrow keys, Home, End, Page Up, Page Down, Tab, and Enter move it; Ctrl/Cmd with an arrow jumps to the edge, and holding Shift extends the block instead of moving it. Ctrl/Cmd+A selects all cells, Ctrl/Cmd+Space the active column, Shift+Space its row—and from there Shift with an arrow key adds whole rows or columns one at a time, while a plain arrow key drops back to a single cell.
• Read a selection's numbers at a glance. Selecting cells, rows, or columns shows Count, Distinct, Sum, Average, Min, and Max in a status bar under the grid—computed in the parser worker, so a block that reaches past the loaded rows is still summarized correctly.
• Read one wide record at a time. Turn on Record view to list the active cell's record as one field per line beside the grid, for a record too wide to read comfortably across. It follows the active cell and the same Up/Down keys as the grid.
• Share results without cleanup. Copy a cell, a rectangular range, selected rows or columns, or the complete table as Excel-compatible TSV or a Markdown table. Use the toolbar, press Ctrl/Cmd+C, or right-click for a copy action labelled for what is selected. Ctrl/Cmd+Shift+C prepends the header row to a cell or row selection. Numeric columns stay right-aligned in Markdown.
• Cut a wide extract down to size. Export… opens a column picker with a filter box, All/None/Invert, and checkboxes that highlight the matching columns in the table as you tick them. Export directly to CSV, TSV, PSV, or Excel; choose the delimiter for custom CSV output and whether to include the header, then save the result as a new file—the source is never touched.
• Send a selection straight to Excel. Switch the export format to Excel workbook (.xlsx) to write a real spreadsheet instead of delimited text: Thai, CJK, and accented text arrive intact with no encoding prompt, the header row is frozen, columns are sized to their content, and clean numeric values become actual numbers you can sum. Identifiers keep their exact glyphs—007 stays 007, and currency and percentage text is never silently rescaled.
• Work where your data lives. Supports Remote - SSH, WSL, Dev Containers, Codespaces, and desktop virtual workspaces such as Remote Repositories.

Your CSV workflow, without leaving VS Code

1. Open a .csv, .tsv, .tab, or .psv file—CSV Sidekick opens it as a table by default.
2. Scan the dataset with a frozen header, filter its full set of records, auto-fit columns, resize fields, or freeze panes at the selected cell. Click a column header to sort ascending, descending, or back into file order, and select Data profile to see each column's quality and distribution summary.
3. Select the values you need: drag across cells for a rectangle, or click a cell, row number, or column header; Ctrl/Cmd-click to add rows or columns; Shift-click to extend a range. Ctrl/Cmd-click and Shift-click on a column header select without re-sorting the table underneath you. Watch the status bar under the grid for the selection's Count, Sum, Average, Min, and Max, or turn on Record view to read one wide record at a time.
4. Copy the selection for Excel or Markdown, use Export… to save the columns you picked as a new CSV, TSV, PSV, or .xlsx file—opening it with a rectangle selected ticks that block's columns and offers its rows as the export scope—or choose Open as text when you need to edit the raw file.

The toolbar buttons are icons—hover one for its name and what it does—and every table action is also available from the right-click menu.

Safe inspection and practical cleanup

The table viewer is read-only: it never modifies the source file. This makes it a safe place to inspect client extracts, experiment outputs, and production data.

For an editable CSV in the text editor, right-click and use 🧰 PK CSV to:

• clean header names or convert them to snake_case, camelCase, PascalCase, or kebab-case;
• trim whitespace from every field;
• remove the trailing delimiter that some exporters leave at the end of every line;
• copy the complete file as Excel-ready data or a Markdown table; and
• toggle the same column-color highlighting used by the table view.

Text-editor cleaning commands modify the document and support normal Undo.

Remote and read-only data

Install CSV Sidekick in the VS Code location offered for your remote target, then open CSV-family files normally or choose Open With…. Viewing and copying work for virtual repositories that are read-only; text-editing commands are unavailable when the file cannot be changed.

The browser-only editors at vscode.dev and github.dev are not currently supported because parsing runs in the Node.js extension host.

Settings

Configure these globally or per workspace, then reopen the file to apply a change.

Contributing

See DEV.md for development and build instructions.