Edit an offline computing task SKILL-阿里云帮助中心

You can use natural language commands to instruct code agents, such as Qoder and CodeX, to modify an existing offline computing task, including its task code, scheduling configuration, data source, and upstream and downstream dependencies. This topic describes the features, prerequisites, and examples of this capability. An appendix also provides a list of input parameters.

Features

This feature lets you modify the configuration of existing offline tasks in the development environment using natural language commands. Common modifications include adjusting task code, modifying the scheduling cycle and cron expression, switching the data source, configuring a task to run at continuous intervals all day, and adjusting the node output name, priority, runtime parameters, and upstream dependency. Before editing, the agent typically locates the task by its name and reads its current configuration as a baseline before applying the requested changes.

Use cases:

Modify the task code or scheduling configuration of an existing offline computing task.
Switch the data source, database name, or catalog for SQL-based tasks.
Change a task's schedule to run at minute-level intervals or adjust its cron expression.
Adjust a task's upstream dependency, node output name, priority, or runtime parameters.

Note

Limitations: Edit operations update the task configuration only in the development environment. The task type cannot be changed. To change the type, you must delete the task and then recreate it. To apply changes to the production environment, you must submit and publish them. Every edit completely overwrites the upstream dependencies. To avoid removing dependencies, you must provide the complete list with each edit.

Prerequisites

The target Dataphin project exists, and you are a member of the project.
The target offline computing task has been created in the development environment and can be found by its name.
Have the new content ready, such as the updated task code, scheduling configuration, or dependency information.
When editing a database SQL task, have the target data source information ready. In projects with dual development and production environments, you typically need to use a production data source.

Examples

Describe your requirements to the agent in natural language. The following are example commands:

"Change the code for this task to select * from demo01."
"Help me change the schedule for this task to run every 5 minutes all day."
"Switch the data source for this SQL task to another PostgreSQL data source."
"Add an upstream dependency to this task that depends on the shell_a task."

Interaction steps:

The agent first locates the task by its name within the project directory. If any information is missing, the agent prompts you for it.
The agent typically reads the task's current configuration as a template and then applies the modifications based on your request.
When editing an upstream dependency, the agent overwrites the entire list with the one you provide. To prevent data loss, always supply the complete dependency list.
After the modifications are complete, the agent informs you that it applied the changes only to the development environment. To deploy them to production, you need to submit and publish the task.

Input parameters

The following table lists the input parameters for editing an offline computing task. The agent automatically identifies required parameters from your commands and prompts for any missing details. You do not need to provide system-generated parameters, which are marked as not required.

Parameter	Required	Description	Example
tenant ID	Yes	The identifier for the Dataphin tenant where the operation is performed.	1234567890123456789
project ID	Yes	The Dataphin project to which the task belongs.	—
Task name	Yes	The name of the offline computing task to edit. The agent uses this name to locate the task.	cli_test
task type	Yes	The type of task. This must match the original `task type` and cannot be changed during an edit.	Shell
task code	No	The new content of the `task code`. Provide this only if you need to modify the code.	select * from demo01
scheduling cycle	Conditionally required	Required when you modify the schedule, for example, daily, hourly, or every minute.	Every minute
cron expression	Conditionally required	Required when you modify the schedule. Defines the specific time when the task is triggered.	Daily at 00:00
data source	Conditionally required	Provide this when you switch the `data source` for an SQL-based task. In projects with dual development and production environments, you typically need to use a production `data source`.	—
Upstream dependency	No	The complete list of the task's upstream dependencies. This is completely overwritten during an edit.	shell_a
priority	No	The scheduling `priority` of the task. Valid values range from 1 to 9.	5