npm / zbsearch

zbsearch

What changed in zbsearch 3 to 4, from its official release notes, upgrade guides and advisories. Sources checked 2026-10-19.

1 breaking13 new APIs and options2 performance5 behavior and defaults5 fixes and other

See which of these apply to your code, ranked, in your agent:

npx @unstale/cli init

zbsearch 422 changes

Splitting keeps letters from other scripts and trims punctuation at token edges.
4.0.1release notes

Engine primitives needed by a lazily-loaded static index are exposed.
4.0.1release notes
A sharded, lazily-fetched index is added for static sites.
4.0.1release notes
Adds the chunkedLoad method.
4.0.0release notes
A Docusaurus plugin is now available.
4.0.0release notes
IndexedDB persistence is now supported.
4.0.0release notes
Adds the insertMultipleAsync method for asynchronous multiple inserts.
4.0.0release notes
Adds the loadAsync method.
4.0.0release notes
S3 persistence is now supported.
4.0.0release notes
Instances can now be created without a schema.
4.0.0release notes
Slovak language support is now available.
4.0.0release notes
Suggestions APIs are now available.
4.0.0release notes
A Vitepress integration is now available.
4.0.0release notes

The docs-index payload is trimmed and hydrated off the main thread.
4.0.1release notes
Document insertion reuses the tokenizer normalization cache.
4.0.1release notes

The English stopword list now includes “a”.
4.0.1release notes
Repeated tokens are kept by default so BM25 sees term frequencies.
4.0.1release notes
The sharded index is shipped through docs-index and every framework plugin.
4.0.1release notes
The markdown parser for docs integrations has been improved.
4.0.0release notes
Search quality has been improved.
4.0.0release notes