sqrail runs read-only SQL over CSV, TSV, JSON, and Parquet.

sqrail schema [--memory SIZE] [--threads N] [--timeout DURATION]
              [--max-input-files N] [--strict-schema] FILE...
sqrail check [-t NAME=PATH]... [--memory SIZE] [--threads N] [--timeout DURATION]
             [--max-rows N] [--max-input-files N] [--max-sql-bytes SIZE]
             [--strict-schema] [SQL|-]
sqrail run [-t NAME=PATH]... [-o FILE] [--memory SIZE] [--threads N]
           [--spill DIR [--max-spill SIZE]] [--timeout DURATION] [--max-rows N]
           [--max-output-bytes SIZE] [--max-input-files N] [--max-sql-bytes SIZE]
           [--stats] [--strict-schema] [SQL|-]
sqrail --version

Names/types result: schema once. Otherwise trust stated names/types, run once, and stop after success.
-t binds a file, Parquet directory, or glob; '-' reads SQL; '--' ends options.
Files union columns by name; --strict-schema requires exact schemas.
check emits columns, inputs, and plan without execution.
No -o: JSONL stdout. -o: extension selects format.
Limits fail closed; --stats writes versioned success JSON to stderr.
Text supports .gz/.zst. Outputs are private, atomic, never overwritten.
SQL is one SELECT, VALUES, or WITH; order needs ORDER BY.
Errors are one stderr JSON. Exit: 0 success, 2 usage, 3 input, 4 SQL, 5 output, 70 internal.