How does the output structure the rows?+
Each CSV row becomes one TOML table in an array, written as `[[rows]]` followed by `key = value` lines for every column. An array of tables is the standard TOML idiom for a list of records, so the result loads cleanly with any TOML parser as an array under the table name.
Can I rename the table?+
Yes. There's a field for the table name (default `rows`), so you can match the key your config or Rust/Go/Python loader expects — for example `[[servers]]` or `[[users]]`. If the name has spaces or punctuation it's automatically quoted.
How are types decided?+
From column type inference. All-numeric columns emit bare numbers via our number parser, true/false-style columns become `true`/`false`, and everything else — including dates — is emitted as a quoted, escaped string. Empty cells become an empty string `""` because TOML has no null.
What about special characters and quotes in my data?+
Strings use TOML basic-string escaping: backslashes, double quotes, tabs, newlines and other control characters are escaped (`\\`, `\"`, `\n`, `\uXXXX`). The output is guaranteed to round-trip through a compliant TOML parser.
Why are dates quoted instead of native TOML datetimes?+
Native TOML datetimes are strict about format, and raw CSV dates are often ambiguous (03/04/2025) or partial. We quote them as strings by default so nothing is silently mis-parsed. Toggle 'Normalize dates to ISO 8601' to rewrite them to a clean ISO string first.