For JSON content, cap arrays that are both long and large. An array with
more than max_items elements whose compact JSON is over max_chars
keeps leading elements up to max_chars (never fewer than max_items)
and ends with a summary string element. Smaller arrays stay whole, so a
search result or a page of records keeps every entry.
For plain text / CLI output, collapse runs of repeated identical lines
down to at most max_repeated_lines.
Config options: max_repeated_lines (u32, default 3).
Non-plain-text content passes through unchanged.
For JSON content, flatten nested objects up to max_depth levels using
dot-notation for flattened keys (e.g. {"a":{"b":1}} → {"a.b":1}).
Config options: max_depth (u32, default 3).
Non-JSON content passes through unchanged.
For git diff output, fold consecutive unchanged context lines (lines
starting with a space) into a compact [N unchanged lines] marker.
This preserves all changed lines (+/-) and hunk headers (@@) while
dramatically reducing noise from context lines.
For JSON content, keep only the specified top-level fields; drop all others.
Config options: fields — array of field name strings.
Non-JSON content passes through unchanged.
For JSON content, remove specified fields by key name.
Supports dot-notation for nested fields (e.g. “metadata.internal_id”).
Config options: fields — array of field path strings.
Non-JSON content passes through unchanged.