> For the complete documentation index, see [llms.txt](https://cortex-docs.paloaltonetworks.com/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://cortex-docs.paloaltonetworks.com/cortex-agentix/configure-cortex-agentix/configure-the-cortex-agentic-assistant/agents-hub/expand-agent-capabilities-with-mcp-integrations.md). # Expand agent capabilities with MCP integrations Connect Model Context Protocol (MCP) integrations in Cortex AgentiX to extend Agentic Assistant agents with third-party tools and external system access. Cortex Agentic Assistant connects agents to external tools with Model Context Protocol (MCP) integrations. Agents use third-party MCP server tools to retrieve data and perform tasks in external systems. For example, an agent can open a Jira issue or check whether security scans are bypassed in GitHub workflows. Cortex Agentic Assistant connects to external MCP servers through streamable HTTP. It supports OAuth-based and authless servers. The MCP server must be accessible through a URL. Install the relevant Cortex Marketplace content pack and configure an MCP integration instance. The integration discovers available tools and automatically generates agentic actions. {% hint style="info" %} For agentic actions to be created from tools on an MCP server, the tools must have input and output descriptions on the MCP server. These descriptions are required for Cortex AgentiX to understand the tools' capabilities and the correct use cases. {% endhint %} ### Configure MCP integrations To find MCP content packs, go to **Settings** → **Marketplace**. Under **Types**, filter for **MCP**. Examples include **Cloudflare MCP**, **GitHub MCP**, and **Atlassian Cloud MCP**. Use the **Generic MCP** content pack for MCP servers without a dedicated content pack. Each MCP integration includes required URL and authentication parameters. You can create multiple integration instances for each MCP integration. For example, you might configure one instance of the **GitHubMCP** integration to connect to an environment with read tools and another instance to connect to an environment with both read and write tools. In addition, the **GenericMCP** integration can be used to connect to multiple MCP servers, each with a separate integration instance. When you **Test** the integration instance, Cortex AgentiX verifies server connectivity. {% hint style="info" %} If you are using OAuth-based authentication, the **Test** button returns an error containing the command to run in the playground in order to test the connection. {% endhint %} All configured MCP integration instances can be viewed in the **Settings** → **Data Sources & Integrations** page. You can view each integration instance and verify the status of the connection, **Test** the connection, enable or disable the integration instance, and view the last discovery timestamp. ### Discover MCP tools and agentic actions The integration instance checks hourly for new or changed third-party MCP server tools. Checks also run whenever you save an integration instance. Discovered tools are automatically registered as AI actions of type **MCP Tool**. Actions are created for each MCP integration instance. Multiple instances for the same MCP server create multiple actions for the same tools. Each action name includes the server, tool, and integration-instance names. Tool discovery creates system actions. You cannot edit these actions. You can enable or disable them. You can also mark actions as sensitive, requiring manual approval. By default, all MCP server actions are sensitive. {% hint style="info" %} * If an MCP tool is removed from the MCP server, the action will be unavailable due to missing content. If the tool is restored on the MCP server, the action is automatically re-enabled. * If you have a development and a production tenant, you can push actions created from MCP tools from development to production. {% endhint %} ### Add MCP tools to agents and manage permissions Add MCP tool actions to custom agents in the **Agentic Assistant Hub**. By default, all users with custom agent access can use its available tools. To restrict MCP tool access, go to **Settings** → **Configurations** → **Data Collection** → **Integration Permissions**. Restrict MCP integration instance commands to one or more roles. Only users in permitted roles can use restricted actions. --- # Agent Instructions This documentation is published with GitBook. GitBook is the documentation platform designed so that both humans and AI agents can read, navigate, and reason over technical content effectively. Learn more at gitbook.com. ## Querying This Documentation If you need additional information that is not directly available in this page, you can query the documentation by asking a question. Perform an HTTP GET request on the following URL with the `ask` and `goal` query parameters: ``` GET https://cortex-docs.paloaltonetworks.com/cortex-agentix/configure-cortex-agentix/configure-the-cortex-agentic-assistant/agents-hub/expand-agent-capabilities-with-mcp-integrations.md?ask=&goal= ``` `ask` is the immediate question: it should be specific, self-contained, and written in natural language. `goal` is what the user is ultimately trying to achieve, the reason they need the answer. Sharing it helps GitBook give you a better, more relevant answer. A goal is most helpful when it describes the outcome the user wants rather than restating the question. For example, with `ask=how do I create an API token`, a goal like `automate deployments from our CI pipeline` lets GitBook tailor the answer to that use case. The response will contain a direct answer to the question and relevant excerpts and sources from the documentation. Use this mechanism when the answer is not explicitly present in the current page, you need clarification or additional context, or you want to retrieve related documentation sections.