add-column-to-catalog

Use the column.js helper with --add-column and a JSON spec containing name, description, and position. It inserts the header cell, separator, legend row, and pads every data row with "—" using a code-span-aware parser instead of regex.

Run column.js with --populate-column and the column name. Known fetchers handle pricing, funding, last_updated, context_window, and offline_capable; other columns fall back to generic homepage and GitHub research, writing "—" when no source is found.

Yes. The splitRow helper is code-span aware, so cells containing `\|` or backtick-wrapped pipes are parsed correctly and not split at the wrong position. New cell values are also escaped or wrapped to keep the table valid.

The operation is idempotent: if the normalized column name already exists, the DDL step is skipped and only population runs. With overwrite set to false, only empty or "—" cells are filled; existing values are preserved.

Yes. Pass --dry-run to either --add-column or --populate-column to report the planned insertion position and row counts without writing the file. git diff --stat will show no changes after a dry run.