Execute SQL statement
Executes an SQL statement and returns the results. Supports both regular and streaming responses. For streaming, use Accept: application/stream+json header.
The database can be specified in the request body, or defaults to the database from the authentication token or system settings.
Streaming Large Results
For queries that may return large result sets, use the streaming response format:
With streaming, results are returned as they become available using HTTP chunked transfer encoding. Each chunk contains a valid JSON object that can be parsed independently.
Parameterized Queries
For queries with variable inputs, use parameterized queries instead of string concatenation:
Authentication
Headers
Supported compression algorithms for responses. Multiple algorithms can be specified as a comma-separated list. Common values include ‘gzip’, ‘deflate’, ‘br’ (Brotli), or combinations like ‘gzip, br’.
Note that for large result sets, compression may produce suboptimal results. Without compression the result sets can be chunks of rows and processed in a streaming fashion. With compression the entire result set must be compressed before it is returned to the client. Thus the entire result set will be decompressed on the client before any rows can be processed
Indicates the preferred encoding type from those specified in Accept-Encoding. This header provides fine-grained control when Accept-Encoding contains multiple values, allowing the client to explicitly specify which encoding is preferred for this specific request.
Specifies the desired compression level for the selected encoding algorithm. Values typically range from 1 (fastest, least compression) to 9 (slowest, maximum compression). The exact behavior depends on the specific compression algorithm being used.
Query parameters
Response format:
- array - Returns schema and data as arrays (more efficient for large datasets)
- collection - Returns data as JSON objects (more convenient for client processing)
Note: This parameter is only applicable when using Content-Type: text/plain. When using Content-Type: application/json, specify format in the request body.
The schema to use for the SQL statement.
Note: This parameter is only applicable when using Content-Type: text/plain. When using Content-Type: application/json, specify schema in the request body.
Request
Parameters to substitute in the SQL statement. The keys should match the parameter names in the statement (without the enclosing braces). String values are wrapped in single quotes, all other types are unquoted. Null values will be expanded as null.
Example: If your SQL is SELECT * FROM orders WHERE order_date > {start_date},
you would provide {"start_date": "2023-01-01"} in this field.
Response format:
- array - Returns schema and data as arrays (more efficient for large datasets)
- collection - Returns data as JSON objects (more convenient for client processing)
Parameters to substitute in the SQL statement. The keys should match the parameter names in the statement (without the enclosing braces). String values are wrapped in single quotes, all other types are unquoted. Null values will be expanded as null.
Example: If your SQL is SELECT * FROM orders WHERE order_date > {start_date},
you would provide {"start_date": "2023-01-01"} in this field.
Number of rows to return in each chunk for streaming responses. A value of 0 means the database can determine how many to return.
- For interactive applications: Use smaller values (100-1000)
- For batch processing: Use larger values (5000-10000)
The maximum number of rows to return. Note that the SQL LIMIT clause is much preferred, since the query engine will generate the total number of rows, but then only max_rows will be returned to the caller.

