|
| 1 | +--- |
| 2 | +title: Get started with custom connectors using AI agent in Microsoft Sentinel |
| 3 | +description: Custom Data connectors using AI agent in Microsoft Sentinel Visual Studio Code extension |
| 4 | +author: EdB-MSFT |
| 5 | +ms.author: edbaynash |
| 6 | +ms.topic: feature-availability |
| 7 | +ms.date: 03/27/2026 |
| 8 | + |
| 9 | +#customer intent: As a security analyst, I want to easily create custom connectors in Microsoft Sentinel using an AI agent, so that I can quickly integrate new data sources without extensive manual effort. |
| 10 | +--- |
| 11 | + |
| 12 | +# Get started with custom connectors using AI agent in Microsoft Sentinel |
| 13 | + |
| 14 | +The Microsoft Sentinel connector builder agent builds data connectors in minutes using the AI‑assisted workflow in GitHub Copilot using the Microsoft Sentinel extension for Visual Studio Code (VS Code). This low‑code experience guides developers and Independent Software Vendors (ISVs) end‑to‑end by autonomously generating schemas, deployment assets, connector UI, secure secret handling, and polling logic. Built‑in validation surfaces any polling issues early, so you can validate event logs before deploying and ingesting data. |
| 15 | + |
| 16 | +The Sentinel Connector builder agent helps you: |
| 17 | + |
| 18 | +- Reduce manual time and effort in authoring Codeless Connector Framework (CCF)–based connectors |
| 19 | + |
| 20 | +- Scaffold connector definitions using simple prompts |
| 21 | + |
| 22 | +- Iterate on connector logic using natural language |
| 23 | + |
| 24 | +- Validate connector artifacts before deployment |
| 25 | + |
| 26 | +## Prerequisites |
| 27 | + |
| 28 | +Before you begin, ensure that you meet the following requirements: |
| 29 | + |
| 30 | +- An active Microsoft Sentinel workspace |
| 31 | + |
| 32 | +- Access to Visual Studio Code with GitHub Copilot |
| 33 | + |
| 34 | +- The [Microsoft Sentinel VS Code extension](https://marketplace.visualstudio.com/items?itemName=ms-security.ms-sentinel) installed |
| 35 | + |
| 36 | +- Microsoft Sentinel [Contributor](./roles.md#built-in-azure-roles-for-microsoft-sentinel) role to create or modify Sentinel data connectors |
| 37 | +- [Claude Sonnet 4.5 or later](https://claude.ai/login) |
| 38 | + |
| 39 | +## Benefits of Connectors with Agentic experience |
| 40 | + |
| 41 | +The Sentinel Connector builder agent can **reduce connector development time from weeks to hours** for many common scenarios. Tasks that previously required multiple tools, manual handoffs, and repeated validation cycles can now be completed inline, enabling faster iteration and quicker readiness for deployment. |
| 42 | + |
| 43 | +|Area|Non-AI connector development process|VS Code extension with Connector Builder Agent| |
| 44 | +| -------- | -------- | -------- | |
| 45 | +|Authoring experience|Connector definitions, schemas, and configuration items are often authored across multiple tools, including the Azure portal, editors, and JSON templates. Context switching is common.|Connector authoring happens directly in **VS Code**, alongside other development assets, using a single, developer‑focused environment.| |
| 46 | +|Iteration speed|Making changes typically require navigating between tools, updating items, and revalidating manually, which slows down iteration.|Developers can **iteratively refine connectors** using the agent within VS Code, reducing friction between design, update, and review. | |
| 47 | +|Validation and feedback|Validation steps are often performed later in the workflow, increasing the risk of discovering schema or configuration issues late.|Validation happens **closer to the authoring experience**, helping identify issues earlier and improving overall quality before deployment.| |
| 48 | +|Developer productivity|Developers spend time managing tooling and navigation rather than focusing on connector logic and correctness.|Developers can focus on reviewing code and structuring schema using the low-code agent driven connector buildout.| |
| 49 | + |
| 50 | +## Create a custom connector using Sentinel connector builder agent |
| 51 | + |
| 52 | +The following steps show how to create, validate, and deploy a custom Microsoft Sentinel connector using the Sentinel Connector Builder agent in VS Code. |
| 53 | + |
| 54 | +### Step 1: Install and open the Sentinel extension for VS Code |
| 55 | + |
| 56 | +1. [Install Microsoft Sentinel extension for Visual Studio Code](/azure/sentinel/datalake/notebooks) and reload VS Code if prompted. |
| 57 | + |
| 58 | +1. Create and open an empty folder in File explorer. All agent-generated files are locally saved in this folder. |
| 59 | + |
| 60 | +### Step 2: Prompt the Sentinel Connector builder agent |
| 61 | + |
| 62 | +1. Open the VS Code chat and set the chat to Agent mode. |
| 63 | + |
| 64 | +1. Prompt the agent using `@sentinel`. When prompted, select `/create-connector` and select any supported API. |
| 65 | + |
| 66 | + For example, enter the prompt as follows: |
| 67 | + |
| 68 | + ``` |
| 69 | + @sentinel /create-connector Create a connector for Contoso. Here are the API docs: https://contoso-security-api.azurewebsites.net/v0101/api-doc |
| 70 | + |
| 71 | + ``` |
| 72 | + |
| 73 | + :::image type="content" source="media/custom-connector-builder-agent/connector-vscode-prompt-agent.png" alt-text="Screenshot of the agent prompt in custom connector in VS Code" lightbox="media/custom-connector-builder-agent/connector-vscode-prompt-agent.png"::: |
| 74 | +
|
| 75 | +1. Provide the source API information, authentication methods to generate the appropriate connector pattern. |
| 76 | +
|
| 77 | + |
| 78 | +### Step 3: Generate or update connector artifacts |
| 79 | +
|
| 80 | +Based on your input, the agent generates the following four files: |
| 81 | +
|
| 82 | +- Polling configuration |
| 83 | +
|
| 84 | +- Data collection rules (DCR) mappings |
| 85 | +
|
| 86 | +- Connector definition |
| 87 | +
|
| 88 | +- Schema and table references aligned with Sentinel requirements |
| 89 | +
|
| 90 | +The illustration shows the JSON connector files generated. |
| 91 | +
|
| 92 | + :::image type="content" source="media/custom-connector-builder-agent/connector-files-sample.png" alt-text="Screenshot of the connector file generated in VS Code" lightbox="media/custom-connector-builder-agent/connector-files-sample.png"::: |
| 93 | +
|
| 94 | +>[!NOTE] |
| 95 | +> During agent evaluation, select **Allow responses once** to approve changes, or select the option **Bypass Approvals** in the chat. It might take up to several minutes for the evaluations to finish. |
| 96 | +
|
| 97 | +Iteratively refine the connector using agent or inline directly into the JSON files generated. For example: |
| 98 | +
|
| 99 | +- Ask the agent to modify description, author name etc. |
| 100 | +
|
| 101 | +- Update ingestion logic for table name |
| 102 | +
|
| 103 | +- Adjust authentication or polling parameters; for example, polling frequency, timeout window, and others |
| 104 | +
|
| 105 | +>[!IMPORTANT] |
| 106 | +> Do not edit or modify the file while it is building. If a field on the file shows an error, it means the build is still in progress. |
| 107 | +
|
| 108 | +### Step 4: Validate the connector configuration |
| 109 | +
|
| 110 | +1. To validate the API for data source events, right click on the folder that contains the ARM template and select **Microsoft Sentinel** > **Test Connector**. |
| 111 | +
|
| 112 | + :::image type="content" source="media/custom-connector-builder-agent/connector-validation.png" alt-text="Screenshot of the validation step for custom connector in VS Code" lightbox="media/custom-connector-builder-agent/connector-validation.png"::: |
| 113 | + |
| 114 | +1. In the **Test Connector** pane, enter the authentication details for your data source API, and then select **Connect**. |
| 115 | +
|
| 116 | + - Polling starts based on the settings in your polling configuration JSON file. |
| 117 | +
|
| 118 | + - On the **Events** tab, review the request headers and the events returned by the API. |
| 119 | +
|
| 120 | + > [!NOTE] |
| 121 | + > This test confirms that the API call succeeded and returns events. It doesn’t confirm that events are being written to your Sentinel table. Table ingestion is validated when you finish setting up the connector in the Data connectors page in Microsoft Sentinel. |
| 122 | + |
| 123 | +After you validate the connection, select **Disconnect** to stop the polling session. |
| 124 | +
|
| 125 | +### Step 5: Deploy |
| 126 | +
|
| 127 | +1. After the validation test succeeds, select **Deploy** in the chat window to start deploying the connector. |
| 128 | +
|
| 129 | + :::image type="content" source="media/custom-connector-builder-agent/connector-deploy.png" alt-text="Screenshot of the deploy step for custom connector in VS Code" lightbox="media/custom-connector-builder-agent/connector-deploy.png"::: |
| 130 | +
|
| 131 | +1. The extension opens a panel where you can choose from your available Microsoft Sentinel workspaces. |
| 132 | +
|
| 133 | +1. Select a workspace, then select **Deploy** to deploy the connector to that workspace. |
| 134 | +
|
| 135 | + :::image type="content" source="media/custom-connector-builder-agent/connector-workspace-deploy.png" alt-text="Screenshot of the deploy step to choose workspace" lightbox="media/custom-connector-builder-agent/connector-workspace-deploy.png"::: |
| 136 | + |
| 137 | +Alternatively, right-click the folder that contains the generated files, select **Microsoft Sentinel** > **Deploy Connector**. |
| 138 | +
|
| 139 | + :::image type="content" source="media/custom-connector-builder-agent/connector-files-deploy.png" alt-text="Screenshot of the alternate way of deploying in VS Code" lightbox="media/custom-connector-builder-agent/connector-files-deploy.png"::: |
| 140 | +
|
| 141 | +When deployment completes, a success message appears in the **Output** window. |
| 142 | +
|
| 143 | + :::image type="content" source="media/custom-connector-builder-agent/successful-deployment.png" alt-text="Screenshot of successful deployment" lightbox="media/custom-connector-builder-agent/successful-deployment.png"::: |
| 144 | + |
| 145 | +> [!NOTE] |
| 146 | +> After deployment into a Sentinel workspace, this connector focuses on ingesting data into Microsoft Sentinel tables. It doesn’t include a complete solution package or prebuilt end-to-end SOAR workflows for security coverage. If you need automation, create the required playbooks and workflows for your scenario. |
| 147 | +
|
| 148 | +## Getting Help |
| 149 | +
|
| 150 | +- For ISV partners building integrations, contact: [[email protected]](mailto:[email protected]) |
| 151 | +- For technical questions, use [Microsoft Q&A](/answers/topics/azure-sentinel.html) with the tag 'azure-sentinel' |
0 commit comments