Limitations

No output format. The result is a list of entries. Writing a unified diff, HTML or colored output is up to you; see Displaying a diff for examples.
No patches or merges. The library does not apply a diff to a text, read a patch file, or merge three versions of a text.
No move detection. A block of text moved elsewhere is a deletion and an insertion.
No work limit and no cancellation. A diff runs to completion however long it takes, and there is no CancellationToken. Diffs of large, unrelated texts can be slow, especially at the character level; see Algorithms.
Ordinal comparisons only. IgnoreCase uses ordinal case-insensitive rules; there is no culture-aware or accent-insensitive comparison.
Characters are UTF-16 code units. TextChunker.Characters splits surrogate pairs and combining sequences into several chunks. See Characters.
Texts are strings. Both texts are read into memory as string; there is no streaming or file API.