Import / Export
Dataworkz allows you to export Agents as JSON files and import them into the same or a different Dataworkz environment. This is useful for sharing Agents across teams, promoting Agents between environments such as development and production, or creating backups of Agent configurations.
Exporting an Agent
Step 1: Open the Agent Actions Menu
From the My Agents page, locate the Agent you want to export. Click the three-dot menu in the Actions column next to the Agent name to open the actions menu.
The following actions are available from this menu:
Edit
Open the Agent configuration for editing
Export
Export the Agent as a JSON file
References
View all references to this Agent across the environment
Insights
View usage and performance insights for the Agent
Delete
Permanently remove the Agent
Step 2: Click Export
Click Export from the actions menu. The Export Agent dialog opens.
Step 3: Review What Is Included
The Export Agent dialog clearly shows what is and is not included in the export:
This export includes:
Agent structure and configuration
Tools and Workflows
Possible setup required after import:
Authentication credentials
Data source connections
Document access
LLM configuration
⚠️ Warning: The export does not include credentials, connection strings, or environment-specific configuration. These must be reconfigured manually after importing the Agent into a new environment.
Step 4: Choose Whether to Include Context Variables
The dialog displays any Context Variables defined for the Agent and allows you to decide whether to include them in the export. Each variable is shown with its name, description, data type, and current value.
For example:
ingestName
String
${myFilesIngestName}
Review each variable and confirm whether it should be included in the exported file.
Step 5: Click Export
Click Export to download the Agent as a JSON file. Click Cancel to close the dialog without exporting.
Importing an Agent
Step 1: Click Import
From the My Agents page, click the Import button in the top right corner. Select the Agent JSON file you exported previously and upload it.
The Import Agent dialog opens, displaying the Agent name, description, and a structured configuration wizard that guides you through the steps required to make the Agent operational in the current environment.
Step 2: Configure Tool Parameters
The import wizard identifies all Tools included in the Agent and flags any that require configuration before the Agent can run. The left panel shows two sections:
Tool Parameters — lists all Tools that need attention, with a count of how many Tools require configuration
Context Variables — lists all context variables that need to be reviewed or updated
Each Tool is listed with its name, type, and description. Tools flagged with a warning icon require configuration — typically because they reference data sources, LLMs, or credentials that do not exist in the current environment and must be set up before the Agent can function.
Expand any Tool to configure its parameters. For example, LLM Tools will prompt you to select an LLM from those available in the current environment:
Available LLMs are shown in a dropdown — for example, gpt35, gpt4, dev_gpt_4o, modal_llama33_70b, Llama_3_2. Select the appropriate model for each Tool that requires one.
The right panel displays the Agent's Notes alongside Tool-level notes, providing context about the Agent's purpose and any setup requirements documented by its creator.
Step 3: Configure Context Variables
Click Context Variables in the left panel to review and update the global context variables for the imported Agent.
Each context variable is displayed with its name, description, data type, and current value. Update any values that differ between the source and target environments — for example, replacing a folder name or ingest reference with the correct value for the current environment.
Step 4: Save or Skip
Once all required Tool parameters and context variables are configured, click Save to complete the import and make the Agent available in your environment.
If you want to import the Agent without completing the configuration immediately, click Skip for now! to save the Agent in an incomplete state and configure the remaining parameters later.
⚠️ Warning: An Agent imported with incomplete Tool configurations will not function correctly until all required parameters are set. Ensure all flagged Tools are configured before running the Agent.
Important Notes
💡 Note: Exported Agents are self-contained JSON files that capture the full Agent structure including Scenarios, Tool definitions, and Workflows. They can be version-controlled, shared with team members, or used as templates for new Agents.
💡 Note: When importing into a new environment, review all Tool configurations carefully. Data source connections, authentication credentials, document access, and LLM selections are environment-specific and must be reconfigured to match the target environment.
💡 Note: Context Variables with invoke-time values that reference environment-specific resources — such as folder names or ingest paths — should be updated during import to reflect the correct values for the target environment.
💡 Note: The References option in the Agent actions menu shows all places across the environment where the Agent is referenced — useful for understanding the impact before deleting or significantly modifying an Agent.
Last updated