Skip to main content
Upcoming release — Zylon 1.58These docs are published ahead of release so you can prepare your API integrations. The release image is not yet available. For installations running 1.57, use the 1.57 documentation.
Omnisearch combines keyword and semantic search across projects, chats, artifacts, and artifact content.

​
Configuration

After changing the configuration, run sudo zylon-cli sync.

​
How language configuration works

languages.keywordSearch affects keyword search only. For every configured language, Zylon builds and queries a language-specific PostgreSQL full-text index. Configure every language used by your projects, chats, and documents so stemming and keyword ranking use the appropriate dictionaries. Semantic artifact-content search follows a separate path: it sends the query to PrivateGPT and uses the enabled embedding model. Therefore, semantic language coverage depends on that embedding model, not on languages.keywordSearch or the multimodal chat model. For multilingual semantic retrieval, enable the multilingual capability for your selected AI preset:
This capability selects the multilingual-e5-large embedding model. See AI preset capabilities.
Do not leave languages.keywordSearch empty. Omnisearch rejects requests when no keyword-search language is configured. Unsupported language codes also cause an error.

​
Supported keyword-search languages

Language codes are case-insensitive. Duplicate entries are removed.

​
Query-length behavior

Search queries must contain at least three characters. omnisearch.maxQueryCharacters changes the upper limit enforced by the backend, the Workspace API, and the Workspace search field.
Keep the default value of 80 unless your users need longer queries. The same configured limit applies to both API requests and the Workspace search field.