> For the complete documentation index, see [llms.txt](https://docs.karini.ai/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://docs.karini.ai/prompt-management/deep-agent-beta/create-deep-agent-prompt.md). # Create Deep Agent Prompt Define the workflow objective, required inputs, expected outputs, and any operational constraints before configuring the prompt. This ensures controlled execution and alignment with system requirements.\ To configure a Deep Agent prompt, follow the structured workflow outlined below: * Navigate to **Prompt Playground** and click **Add New**. * From the **Task** dropdown, choose **Deep Agent(Beta)** to enable Deep Agent orchestration and file system capabilities. * Enter a unique and descriptive name for the Deep Agent prompt. * Specify the **Max State Updates** value to control the number of refinement cycles allowed during execution (default: 3). * Write clear and structured instructions in natural language, defining the expected reasoning behavior, tool usage guidance (if applicable), and output format requirements. * Add required variables as needed: * Use `{Variable}` for dynamic input parameters. * Use `{{ ... }}` when including JSON content to prevent variable parsing. * Add the mandatory variable **File System Instructions**, clearly defining: * What information should be stored * Where it should be stored * The format in which it should be stored * Provide the **Agent Input** that the Deep Agent will process. * Click **Save** to complete the configuration. * After saving the agent prompts, the **Tool** tab becomes accessible. Navigate to the **Tool** tab to configure the required settings. * After saving the Deep Agent prompt, the **Tools** tab becomes available. Open the **Tools** tab to configure the required tool settings, and refer to [**Configuring Agent Tools** ](/prompt-management/agentic-prompts/create-agent-prompt.md#configuring-agent-tools)for the standard tool setup steps. Beyond the standard tool configuration capabilities supported by Agent, **Deep Agent** introduces an additional execution control, **Interrupt Tool**, available as a checkbox within the tool configuration panel .This setting provides enhanced governance by enabling pre-execution interception of tool invocations within the Deep Agent workflow. Finally, click **Save** to persist the tool with the configured settings.