> For the complete documentation index, see [llms.txt](https://docs.buzzy.buzz/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://docs.buzzy.buzz/the-building-blocks/mcp/buzzy-custom-mcp.md). # Buzzy Custom MCP Buzzy Custom MCP is the workflow for enabling MCP on an app you built with Buzzy. Once enabled, MCP-aware AI assistants can interact with that app's data and actions through Buzzy-generated tools. Use this section when you want an app to be available to ChatGPT, Claude, or another MCP-compatible client. If you want an AI agent to build or edit the app itself, use [Buzzy Builder MCP](/the-building-blocks/mcp/buzzy-builder-mcp.md) instead. Custom MCP is not a starting point for creating a new app. Enable it after the app exists, the datamodel is clear, and permissions have been reviewed. For a side-by-side comparison of Custom MCP, REST API, Async API, Node.js client, Buzzy Functions, and code widgets, see [API and Extension Surfaces](/developing-and-extending-buzzy/api-and-extension-surfaces.md). ## Video Overview {% embed url="" %} ## What Buzzy Exposes Buzzy generates MCP capabilities from the app you have already built: * datatable list, get, create, update, and delete tools * Buzzy Functions that are available as tools * widgets that can display app data inside an AI assistant experience * OAuth-based user authentication * the same server-side data access rules used by runtime and REST API calls ## Example App-Specific Tools Custom MCP tools are generated from a specific Buzzy app, so the useful tool names and data shapes depend on that app's datamodel. An assistant does not need generic database access; it can use governed tools that match the business objects in the app. | App | Example assistant request | What the MCP client can do through Buzzy Custom MCP | | ------------------------ | -------------------------------------------------------------------------------------------------------- | -------------------------------------------------------------------------------------------------------------------- | | Short Stays property app | "Find available stays in Byron Bay for next weekend and shortlist the three best options for a family." | List and filter property, availability, pricing, and booking-related rows the user can access. | | Short Stays property app | "Create a draft booking enquiry for this guest and attach these preferences." | Create a booking enquiry row and update linked guest preference fields, subject to app permissions. | | Care coordination app | "Show me clients with overdue care plan reviews and create follow-up tasks for their coordinators." | Read client and care plan rows, then create task rows assigned to the right team members. | | Compliance workflow app | "Summarize open high-risk findings and mark the reviewed ones as ready for manager approval." | List findings, update review status fields, and call an exposed Buzzy Function for summary generation if configured. | | Support ticket app | "Update the priority for tickets from Acme Corp and draft a response for each unresolved billing issue." | Search ticket rows, update priority/status fields, and optionally call a response-drafting Buzzy Function. | | Field service app | "Reschedule tomorrow's maintenance jobs for technicians affected by the outage." | Read jobs, technicians, and schedule rows, then update appointment records within the user's access rules. | These examples are intentionally app-specific. A different Buzzy app exposes a different set of CRUD tools, function tools, and allowed fields because the MCP surface follows that app's datamodel and permissions. ## Security Model MCP tools follow the same server-side access and [Private Data](/the-building-blocks/datatables-fields-and-data/private-data.md) rules as the Buzzy runtime and REST API. * MCP calls authenticate as a Buzzy user. * Row access, Field view, and Field edit rules still apply. * Basic Private Data is masked or redacted when the caller should not receive the raw value. * Sensitive Private Data is hidden by default unless the caller has permission to access it through the relevant tool path. * Private Data image and file fields do not expose denied attachment metadata or signed URLs. When enabling MCP for a compliance-oriented app, test with both allowed and denied users. Do not assume an AI assistant should receive more data than the same user would see in the Buzzy app. ## Learn More {% content-ref url="/pages/pLhiHZxENmWpzeMKMOkj" %} [Implementation](/the-building-blocks/mcp/buzzy-custom-mcp/implementation.md) {% endcontent-ref %} --- # 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 dynamically by asking a question. Perform an HTTP GET request on the current page URL with the `ask` query parameter, and the optional `goal` query parameter: ``` GET https://docs.buzzy.buzz/the-building-blocks/mcp/buzzy-custom-mcp.md?ask=&goal= ``` `ask` is the immediate question: it should be specific, self-contained, and written in natural language. `goal` is optional and describes the broader end goal you are ultimately trying to accomplish on behalf of the user. GitBook uses it to tailor the answer towards what is most useful for that goal. 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.