> ## Documentation Index > Fetch the complete documentation index at: https://docs.ycloud.com/llms.txt > Use this file to discover all available pages before exploring further. # Data connectors > Connect external services through Integrations, MCP, or Custom API and use their tools in AI Agent Skills. Data connectors give AI Agents access to tools from external services. Configure a connection once, then add the specific tools an Agent needs to its Skills. Open **AI Agent > Data connectors** to view and manage your connections. ![Data connectors page showing the Integrations, MCP, and Custom API tabs and the available integrations.](https://mintcdn.com/lchnan/RNoD5r5OXiTray6k/product-assets/english-help-2026-10-30/data-connectors-integrations.png?fit=max\&auto=format\&n=RNoD5r5OXiTray6k\&q=85\&s=f70d467e86e5027f147b3b063f161a36) Open AI Agent > Data connectors to view Integrations, MCP servers, and Custom API connectors. ## Choose a connector type The Data connectors page provides three ways to connect an external service: | Type | Use it to | | - | - | | **Integrations** | Connect a supported service through a guided authorization flow. For example, Shopify Admin provides read-only tools for orders, shipment tracking, returns, and refunds. | | **MCP** | Connect an MCP server and make the tools exposed by that server available to your AI Agents. | | **Custom API** | Create a custom HTTP request to an HTTPS endpoint when you need to connect your own API. | Select the corresponding tab to search existing connections or configure a new one. ## Add connector tools to an AI Agent Skill Connecting a service does not automatically add every available tool to every AI Agent. Choose the tools each Agent can use in its Skills. 1. Open the AI Agent you want to configure. 2. In **Profile**, find **Skills** and select **Manage**. 3. Create a Skill or open an existing one. 4. In **Step-by-Step instructions**, type **/**. 5. Expand the relevant **Integration**, **MCP**, or **Data connectors** group, then select the tool you want to insert. 6. Explain when to use the tool, which information to collect, and what to do with its result. 7. Save the Skill, test it in Preview, and publish the AI Agent when it is ready. ![AI Agent Skill editor showing the slash command menu with Shopify Admin tools.](https://mintcdn.com/lchnan/RNoD5r5OXiTray6k/product-assets/english-help-2026-10-30/shopify-skill-command-menu.png?fit=max\&auto=format\&n=RNoD5r5OXiTray6k\&q=85\&s=beb0f5f6d2938e563088e177eca982c4) Type / in Step-by-Step instructions, then select a tool from the connected Shopify Admin integration. > [!TIP] > > Give each connector step a clear success path, an empty-result path, and a failure path. Do not ask the AI Agent to infer information that the tool does not return. For more guidance on triggers, instructions, and testing, see [AI Skills](https://docs.ycloud.com/en/documentation/agents/ai-agent/ai-skills.md). ## Example: Look up a Shopify order This example creates a Skill that can answer questions about a customer's Shopify order and shipment. ### 1. Connect Shopify Admin 1. Go to **AI Agent > Data connectors**. 2. On the **Integrations** tab, open **Shopify Admin** and complete the authorization flow. 3. Confirm that the integration shows **Connected**. Shopify Admin provides read-only tools such as: * `find_orders` to find an order from the information supplied by the customer * `get_order` to retrieve the selected order * `get_order_fulfillments` to retrieve shipment and fulfillment details * `get_fulfillment_events` to retrieve shipment-tracking events * `get_order_returns` and `get_return` to retrieve return details * `get_order_refunds` and `get_refund` to retrieve refund details ### 2. Create the Skill In the AI Agent's **Profile**, select **Manage** under **Skills**, then create a Skill with the following setup. **Name** > Look up a Shopify order **Start trigger** > Use this Skill when a customer asks about the status of a Shopify order, shipment, return, or refund. Do not use it for product recommendations, checkout questions, or requests to modify an order. **Step-by-Step instructions** 1. Ask the customer for the information required to identify the order. 2. Type **/** and insert `find_orders` from **Integration ยท Shopify Admin**. Use it to find the matching order. 3. If no order or more than one order matches, ask the customer for additional identifying information. Do not guess which order they mean. 4. Insert and use `get_order` for the matching order. 5. If the customer asks about shipping, insert and use `get_order_fulfillments`. Use `get_fulfillment_events` when tracking-event details are needed. 6. If the customer asks about a return or refund, insert the corresponding return or refund tool. 7. Summarize only the information returned by the tools. If a tool fails or the requested information is unavailable, explain that you cannot confirm the result and hand off when appropriate. > [!NOTE] > > The tool names above appear in the **/** command menu after Shopify Admin is connected. Select the commands from that menu instead of entering their names as plain text. ### 3. Test before publishing In Preview, test at least these cases: * One order matches the customer's information * No order matches * Multiple orders match * Fulfillment or tracking details are unavailable * The Shopify tool returns an error * The customer asks the Agent to change an order, which the read-only Shopify Admin integration cannot do ## Create a Custom API connector Use a Custom API connector when you want an AI Agent Skill to call your own HTTPS API. On the **Custom API** tab, select **New data connector**. You can also search existing Custom API connectors by name and filter them by status. ![Full Data connectors page with search, status filters, existing connectors, and New data connector.](https://mintcdn.com/lchnan/6u8ntRayC38GXyQF/product-assets/english-help-context-2026-09-25/final-connector-list-annotated.svg?fit=max\&auto=format\&n=6u8ntRayC38GXyQF\&q=85\&s=1e69a488687c963b905ed5916ebd9b9c) Open AI Agent > Data connectors, then select New data connector. ### Name and description Give the connector a name that describes its function, such as `retrieve_order`. In the description, explain when the connector should be used and what it returns. ![Data connector About step with name and description fields highlighted.](https://mintcdn.com/lchnan/gXEJIQXV2JH2VUQJ/product-assets/english-help-2026-09-22/connector-about-annotated.svg?fit=max\&auto=format\&n=gXEJIQXV2JH2VUQJ\&q=85\&s=a57c6777577f87f55494f26d51d28eb2) Describe the connector's purpose in About before configuring its API request. This draft has not been saved. ### Configure the API request Enter the HTTPS endpoint and select the request method: * **GET** reads information. * **POST** creates information. * **PUT** replaces or updates information. * **PATCH** updates part of a resource. * **DELETE** removes information. ![Data connector API connection step with HTTP method and URL highlighted.](https://mintcdn.com/lchnan/gXEJIQXV2JH2VUQJ/product-assets/english-help-2026-09-22/connector-api-annotated.svg?fit=max\&auto=format\&n=gXEJIQXV2JH2VUQJ\&q=85\&s=5effb08c414cdb9ecc3bce86ed58e377) Choose the HTTP method and enter the endpoint URL. This incomplete draft has not sent a request. #### HTTP headers Select **Add header**, then enter each required header as a key-value pair. ![Full API connection form showing Content-Type and Accept request headers.](https://mintcdn.com/lchnan/6u8ntRayC38GXyQF/product-assets/english-help-context-2026-09-25/final-connector-headers-annotated.svg?fit=max\&auto=format\&n=6u8ntRayC38GXyQF\&q=85\&s=f23c6cb0a8f850aa38a904ed0d1900cf) Select Add header and enter each header as a key-value pair. This example has not been saved or sent. #### Request body For requests that send data, enter a valid JSON request body containing the fields required by your API. ![Full data connector page scrolled to the POST JSON request body.](https://mintcdn.com/lchnan/6u8ntRayC38GXyQF/product-assets/english-help-context-2026-09-25/final-connector-body-annotated.svg?fit=max\&auto=format\&n=6u8ntRayC38GXyQF\&q=85\&s=5099ae99db2d135e6253c53ecefb9937) Enter valid JSON in Request Body. This fictional order example has not been sent. ### Test the response Select **Test request** to verify that the connector returns the expected response. > [!WARNING] > > **Testing sends a real API request.** A POST, PUT, PATCH, or DELETE request can create, change, or remove data. When possible, test against a safe environment or begin with a read-only GET request. ![Test response step with Test Request and a disabled Set live button.](https://mintcdn.com/lchnan/6u8ntRayC38GXyQF/product-assets/english-help-context-2026-09-25/final-connector-test-entry-annotated.svg?fit=max\&auto=format\&n=6u8ntRayC38GXyQF\&q=85\&s=087754abdc3b082833f41e0822a3261c) Select Test Request to run the configured API call. No request has been run in this example. ### Set the connector live After a successful test, select **Set live**. Live Custom API connectors can be selected from the **/** command menu in Skill instructions. To stop using a Custom API connector, open it and select **Set to draft**, or turn off its status from the connector list. Review and resolve any listed dependencies before moving it to draft. ![Full Data connectors list with the Status switches for existing live connectors.](https://mintcdn.com/lchnan/6u8ntRayC38GXyQF/product-assets/english-help-context-2026-09-25/final-connector-status-annotated.svg?fit=max\&auto=format\&n=6u8ntRayC38GXyQF\&q=85\&s=36fb2456c31e1e068179cc06e978feed) Use the Status switch in the list to move a live connector to draft. The switches shown here have not been changed. This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.