Google Cloud documentation says Knowledge Catalog now includes Preview support for a remote Model Context Protocol server, allowing AI applications and external LLM clients to discover data assets, search metadata, and retrieve details about catalog entries.
Google Cloud has documented Preview support for a remote Model Context Protocol server in Knowledge Catalog, expanding how AI applications and external LLM clients can interact with cataloged data products and metadata.
Google Cloud’s release notes state that Knowledge Catalog remote MCP server support lets AI agents and data applications create and discover data products, as well as inspect data product metadata, from external IDEs and LLM clients.
A separate Knowledge Catalog release note from Google Cloud Documentation dated May 25, 2026 says Knowledge Catalog data products are generally available, while remote Model Context Protocol server support is available in Preview.
Google’s documentation describes the remote MCP server as a way to connect AI applications such as Gemini CLI, ChatGPT, Claude, and custom applications to Knowledge Catalog. According to the documentation, supported uses include discovering data assets, searching metadata, and retrieving entry details.
Model Context Protocol, or MCP, is designed to give AI tools a structured way to connect with external systems. In this case, Google Cloud is applying that pattern to Knowledge Catalog, a Dataplex capability used to organize and manage metadata about data assets and data products.
The practical effect, based on Google’s documentation, is that supported AI clients can query Knowledge Catalog through a remote MCP server rather than relying only on manual browsing or custom integrations. Google specifically describes use from external IDEs and LLM clients, which suggests the feature is intended for workflows where developers, analysts, or data teams want metadata access from the tools they already use.
Google’s source material does not claim that the Preview feature automates governance decisions or replaces existing data management processes. The documented capabilities are narrower: creating and discovering data products, inspecting metadata, searching metadata, and retrieving entry details.
Google Cloud Documentation identifies the Knowledge Catalog remote MCP server as a Preview feature. In Google Cloud product terminology, Preview features may be subject to change and are not the same as generally available services.
That distinction matters because the same May 25, 2026 Knowledge Catalog release note says Knowledge Catalog data products are generally available, while remote MCP server support remains in Preview. Organizations evaluating the feature should therefore treat the underlying data products capability and the MCP integration as being at different release stages.
Google’s remote MCP documentation names several example clients, including Gemini CLI, ChatGPT, Claude, and custom applications. The documentation frames the feature around AI applications connecting to Knowledge Catalog to discover assets and retrieve metadata. It does not, in the cited excerpts, provide performance claims, adoption figures, or customer examples.
For data teams, metadata access is often a prerequisite for understanding whether a dataset is relevant, trustworthy, or usable in a given project. Google Cloud’s Knowledge Catalog remote MCP server support places that metadata closer to AI-assisted development and analysis environments.
The feature also reflects a broader movement among cloud and software providers to make enterprise data systems accessible through standardized interfaces for AI tools. However, the available Google Cloud sources support only a specific conclusion here: Knowledge Catalog now has Preview remote MCP server support for connecting AI applications and external LLM clients to catalog metadata and data product discovery functions.
Because the feature is documented as Preview, production planning will likely depend on Google Cloud’s Preview terms, organizational risk tolerance, and the details in Google’s implementation guide. The currently cited Google Cloud documentation establishes the feature’s availability, its intended client connections, and its core metadata and data product discovery use cases.
Google Cloud has documented Preview support for a remote Model Context Protocol server in Knowledge Catalog, expanding how AI applications and external LLM clients can interact with cataloged data products and metadata.
Google’s documentation describes the remote MCP server as a way to connect AI applications such as Gemini CLI, ChatGPT, Claude, and custom applications to Knowledge Catalog.
According to the documentation, supported uses include discovering data assets, searching metadata, and retrieving entry details.
Continue reading
