Examples

Choose a recipe for the task you want to complete. Run it in a test database, check the result, then adapt the table names and values.

1. 01
   Choose a taskFind a recipe for rows, text, vectors or relationships.
2. 02
   Load its sampleCreate the tables and records that the recipe names.
3. 03
   Run one exampleUse the connection and request format shown on that page.
4. 04
   Check the resultCompare the returned fields and IDs with the expected outcome.

New to the database?Copy Anchor Link

Start with one small customer table. That page includes its definition, three rows to load and the query result. Then try an exact filter or join the customers to their orders.

For search, follow one complete pair: index text then find a word, or store a vector then search by similarity.

Choose a taskCopy Anchor Link

• Read and change rows — Filter, join, summarize and update structured data.
• Search by similarity — Store embeddings, retrieve matches and inspect vector indexes.
• Search text — Match words, rank documents and summarize results.
• Follow connections — Read relationships, find paths and analyze groups.
• Use several query models — Coordinate separate writes with shared record IDs.
• Handle an error — Understand the response and choose a safe recovery step.

Connection and SDK versionsCopy Anchor Link

These recipes retain their original request and response examples. HTTP snippets use explicit tenant routes. Use the connection address and authentication instructions for your deployment; do not construct a new address by guessing tenant segments.

Language tabs use earlier client interfaces. Check the SDK reference for the version you have installed and the methods it provides. The 1.0 source candidate is public on GitHub, but its packages have not been published to registries. Use the installation instructions for the pinned source version; a registry install may provide an earlier API.

For writes, check each endpoint’s transaction and retry rules. A timeout may leave the result unknown; confirm the outcome before repeating a write.

Read the examples correctlyCopy Anchor Link

• A schema defines the names and types of a table's fields. Sample rows are the small dataset loaded to demonstrate a request.
• The SQL samples use different table definitions. Use a separate test database for an incompatible sample rather than replacing a table that contains useful data.
• A request containing ... or a comment inside an array is shortened for explanation. Replace it with real values before sending it; JSON does not allow comments.
• Responses are examples of the fields to inspect. Counts, scores and IDs depend on the data you loaded; a page calls out when its small sample dataset reproduces them exactly.
• cURL examples use Bash syntax. Run them in a Bash-compatible shell, or translate quoting and environment variables for your shell.

Configuration-specific examplesCopy Anchor Link

• Elasticsearch API recipes — Supported request forms for dedicated deployments.
• Natural-language endpoint recipes — Questions and plan inspection where Ask is enabled; this is not a general SDK capability.
• Existing MySQL listeners — Historical protocol examples with explicit credential limits.
• Existing SQL Server listeners — Preview protocol examples and dated driver checks.

MySQL and SQL Server are not offered as new connection methods. Their examples require an existing enabled listener. The current PostgreSQL database-login path is read-only.

For request fields rather than recipes, use the API reference. For data definitions, start with schemas.