ExeSQL

The ExeSQL component connects to a relational database and executes SQL queries. It is the core component used by Database Agents to translate natural language into SQL and return structured results.

Scenarios

Use ExeSQL when your agent needs to:

• Query a MySQL, PostgreSQL, or MariaDB database in response to user questions
• Execute generated SQL and return results as a formatted table
• Auto-repair failed SQL queries through a configurable retry loop

Configuration

Parameter	Type	Default	Description
Database Type	string	__	Database type: mysql, postgresql, mariadb, MsSQL, BigQuery, Snowflake, SQLite
Host	string	__	Database server hostname or IP address.
Port	number	__	Database server port.
Database	string	__	Name of the database to connect to.
Username	string	__	Database login username.
Password	string	__	Database login password.
Loop	number	3	Maximum number of retry attempts if a query fails.
Top N	number	30	Maximum number of rows to return.
Model	string	—	The language model used for SQL regeneration or auto-repair.
Temperature	number	0.1	Controls randomness in SQL regeneration.

Note

Options might change according to the database type you select. For example, if you choose BigQuery, you will see additional fields for Project ID and Dataset.

Important

The component also include SQL Regeneration Prompt, which is used for the auto-repair mechanism. Do not modify this prompt unless you have a specific reason to do so, as they are designed to optimize the model’s ability to fix SQL errors.

Usage

1. Add the ExeSQL component to your canvas.

2. Connect a Generate component (or Retrieval + Generate) upstream to produce the SQL query.

3. Fill in the database connection details (host, port, database name, username, password).

4. Set loop to the number of auto-repair attempts allowed on query failure.

5. Connect the ExeSQL output to a downstream Generate component to format the results as a human-readable response.