Skip to content

Settings

All settings live under the csv. prefix and can be set globally, per workspace, or per language with a [csv] block.

The quickest way to change any of them is the CSV icon in the activity bar. Its Settings view groups every option under Parsing, Grid, Text mode and Pipelines, and shows each one’s current value beside its name.

  • Click a setting to change it. Booleans toggle immediately; choices open a picker; numbers open a prompt that validates the range.
  • A setting explicitly set in the current scope gets a discard action to reset it to the default. Reset All Settings in the view menu clears every one at once.
  • The control in the view title chooses whether edits go to User or Workspace settings. The active scope is shown next to the view name, and defaults to Workspace when a folder is open.
  • Open in Settings Editor opens VS Code’s own settings UI filtered to this extension.

When a delimited file is active, the view gains a This file group at the top showing its header-row and delimiter overrides. These are the per-file choices made from the grid toolbar, and they take precedence over everything below. Each can be changed or returned to following the settings, and the group has an action to clear both at once.

Default true. Treat the first row as column names. The toolbar checkbox overrides this for a single file, and that override is remembered.

Default auto. One of auto, ,, \t, ; or |. With auto, the delimiter is detected by scoring each candidate on how consistently it splits the first records, with the file extension as a tie-breaker: .tsv and .tab prefer tab, .psv prefers pipe. The toolbar dropdown overrides this per file.

Default 100000. How many rows are loaded into the grid view. SQL still queries the whole file. Cell edits, find and statistics apply to the loaded rows. Whole-table rewrites run on the host against the full file. Raising this costs memory and initial render time.

Default 5000. Maximum rows returned to the SQL results panel. The query still runs over the whole file and the panel reports the true total when a result is truncated.

Default true. Give each column a distinct colour when a CSV file is open in the text editor. Colours come from your theme’s semantic token colours, so they follow the active theme.

Default 20000. Only colour the first N lines, which keeps very large files responsive.

Default true. Report rows whose field count differs from the header row as warnings in the Problems panel. A mismatch usually means an unescaped quote or delimiter. Multi-line quoted fields are handled correctly and are not flagged.

Default .vscode/csv-pipelines. Where the pipeline builder proposes to save new pipelines. Any *.csvpipe.json file anywhere in the workspace is discovered regardless of this setting.

Default true. Allow steps that evaluate code: filter, compute, script, scriptFile, sql and command. These never run in an untrusted workspace whatever this is set to. Setting it to false disables them everywhere; no-code steps continue to work.

Default 10000. Time limit for expression and script steps. External commands get six times this value.

Default 1000. Rows shown in the pipeline preview grid. Does not affect what is written when the pipeline runs for real.

Setting Why it matters
workbench.editorAssociations Set "*.csv": "csv.gridEditor" (and the same for .tsv / .tab / .psv) to open files in the grid by default; use "default" to keep the text editor
files.encoding The grid reads and writes with the encoding VS Code uses for the document
files.autoSave With afterDelay, grid edits save automatically, and onSave pipelines fire accordingly
editor.semanticHighlighting.enabled Required for rainbow columns; the extension enables it for csv and tsv by default
{
"csv.delimiter": "auto",
"csv.maxRows": 250000,
"csv.pipelines.allowScripts": true,
"[csv]": {
"editor.semanticHighlighting.enabled": true
},
"workbench.editorAssociations": {
"*.csv": "csv.gridEditor",
"*.tsv": "csv.gridEditor",
"*.tab": "csv.gridEditor",
"*.psv": "csv.gridEditor"
}
}