> For the complete documentation index, see [llms.txt](https://docs.carto.com/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://docs.carto.com/whats-new/q3-2026.md). # Q3 2026 {% updates %} {% update date="2026-07-24" tags="new,workflows" %} ## Test your MCP Tools before agents call them You can now test a Workflow-backed **MCP Tool** directly from the canvas, running it exactly as an AI Agent would — before you enable it or wire it into an agent. Open the **Run** dropdown and choose **Run MCP Tool test** to execute the tool through the same CARTO MCP Server path a real agent uses. Enter test values in the input parameters popover (pre-filled from your workflow variables), then watch the run stream step by step in the Messages tab. For **Sync** tools you get the response immediately; for **Async** tools you see the full job submission, status polling, and result retrieval an agent goes through. The response is rendered as expandable JSON with row count and execution time, so you can preview the exact output, understand timing, and copy it to write better tool descriptions and agent instructions. A session-scoped Run history keeps every test run, including failures, so debugging no longer means guessing from a live agent. Learn more in the [Workflows as MCP Tools](/carto-user-manual/workflows/workflows-as-mcp-tools.md#test-your-mcp-tool) documentation. {% embed url="" %} {% endupdate %} {% update date="2026-07-21" tags="new,builder" %} ## Give your AI Agents spatial and business context with semantic models You can now add a **semantic model** to an AI Agent, a structured layer of business meaning over your data. It tells the Agent what your tables and columns mean, how your datasets relate, how your metrics are calculated, and how your spatial data is structured, so the Agent writes correct SQL and applies your definitions consistently. It builds on the open [Apache Ossie](https://ossie.apache.org/) standard with a CARTO extension for geospatial fields. You can write it in Builder, import an existing YAML, or let the Agent Config Assistant generate it from your data. Learn more in our [Semantic model documentation](/carto-user-manual/ai-agents/semantic-model.md).
{% endupdate %} {% update date="2026-07-20" tags="new,workflows" %} ## Revamped Import/Export components in Workflows Moving data in and out of Workflows now works across **BigQuery, Snowflake, Redshift, Databricks and Oracle** from the same canvas: * **Import from URL** and **Import from URL using table** now run on Databricks and Oracle, in addition to BigQuery, Snowflake and Redshift. * **Export as File** is a new component that exports your table to a presigned download link, with no bucket configuration required, on all of these providers. * **Export Tables as PDF** turns your workflow results into a shareable PDF report — with a custom title, descriptions and page orientation — on all of these providers. * **Send by Email** now works across all of these providers too, and can attach a download link to the message. The previous BigQuery-only export components remain available as **Export to Bucket (Legacy)** and **Send by Email (Legacy)** for your existing workflows. Under the hood, this adds new **import** and **export** modules to the CARTO Analytics Toolbox for [Databricks](/data-and-analysis/analytics-toolbox-for-databricks/reference/import.md) and [Oracle](/data-and-analysis/analytics-toolbox-for-oracle/sql-reference/import.md). Learn more in the [Input / Output components](/carto-user-manual/workflows/components/input-output.md) documentation. {% embed url="" %} {% endupdate %} {% update date="2026-07-20" tags="new,platform" %} ## Import and export geospatial data on Databricks and Oracle You can now import files into your warehouse and export data back out on **Databricks** and **Oracle** connections, just like you already do on BigQuery, Snowflake, Redshift and PostgreSQL. Bring data in from a local file or a URL in Data Explorer, and export your results from Builder and Workflows in the geospatial formats you work with, including CSV, GeoJSON, Shapefile, GeoPackage, KML, GeoParquet and more. This is powered by our new Import/Export engine, which brings consistent format support across providers with no manual conversion required. Learn more in the [Importing data](/carto-user-manual/data-explorer/importing-data.md) and [Exporting data](/carto-user-manual/maps/exporting-data.md) documentation.
{% endupdate %} {% update date="2026-07-16" tags="new,builder" %} ## Filter your data by clicking directly on the map You can now filter your Builder map by clicking directly on a feature, alongside widgets, parameters, and the selection tool. Filter just the selected feature or, for polygons, H3, and Quadbin cells, use the clicked feature as a spatial mask to filter all intersecting layers at once. Click-based filters automatically propagate to widgets, PDF exports, and viewport data exports, just like any other filter. Click a postcode or region to instantly update population and other metrics in your widgets, or click individual stores, assets, or sensors to isolate and export only those features. Learn more in our [Interactions documentation](/carto-user-manual/maps/interactions.md#filter-by-feature).
{% endupdate %} {% update date="2026-07-14" tags="new,builder" %} ## Manage your organization's custom markers in one place Custom markers have always been available in Builder, but Editors had to upload and manage them map by map. Now Admins can manage a shared set of custom point markers for the whole organization from **Settings > Customizations > Markers**, and Editors pick from them in Builder without re-uploading the same icons into every map. Replace or rename a marker once and the change flows to every map that uses it, so branding stays consistent as your icons evolve. Learn more in the [Configuring custom markers](/carto-user-manual/settings/customizations/configuring-custom-markers.md) documentation.
{% endupdate %} {% update date="2026-07-06" tags="new,builder" %} ## More export formats and controls in Builder Data exports in Builder are now more flexible and more governed. Beyond CSV, you can export to **GeoJSON**, **Shapefile**, **GeoParquet** and **KML**, so you can take a filtered dataset straight into the tools you already use without any manual conversion. As an Editor, you stay in control of what leaves your map: choose which sources are exportable, whitelist the columns that can be exported so sensitive attributes stay in your warehouse, and set the formats available per source. Every export honors the current map state, so filters, SQL parameters and the viewport are all respected. Learn more in our [Exporting data documentation](/carto-user-manual/maps/exporting-data.md).
{% endupdate %} {% update date="2026-07-03" tags="new,workspace" %} ## Organize your maps and workflows with projects and folders You can now organize your maps and workflows into **projects and folders** instead of one flat list. Group related work together, nest folders as deep as you need, and move or create maps and workflows right where they belong. Projects keep your Workspace tidy and make collaboration easier: share a project or folder once and everything inside inherits those access permissions. You set permissions in one place instead of doing it asset by asset. When an asset belongs in more than one project, add a shortcut instead of moving it. Learn more in the [Projects](/carto-user-manual/projects.md) documentation.
{% endupdate %} {% update date="2026-07-03" tags="new,platform" %} ## More flexible sharing for maps and workflows Sharing maps and workflows is now more flexible. Share assets with specific people or groups, with your whole organization, or publicly, and switch between those modes in a single click. Switching keeps the people and groups you already added with the same access permissions, so you never set anything up twice. Projects and folders share the same flexiblity: share once, and every asset inside a project or folder inherits that access. Learn more in the [map sharing](/carto-user-manual/maps/sharing-and-collaboration.md) and [workflow sharing](/carto-user-manual/workflows/sharing-workflows.md) documentation.
{% endupdate %} {% update date="2026-07-02" tags="builder,improvement" %} ## Single-select mode for text parameters Text parameters in Builder can now be set to **single value** instead of always allowing multiple selections. Viewers pick exactly one option from a radio list, and you decide which value is selected by default. Single-select fits any map that should answer for one thing at a time: a single product line, region, store, audience segment, or scenario in a what-if comparison. The viewer picks one option and the map returns one clear, correct result. Learn more in the [Text parameter](/carto-user-manual/maps/sql-parameters/text-parameter.md) documentation.
{% endupdate %} {% update date="2026-07-01" tags="builder,improvement" %} ## Pie and Time Series widgets now sync with your layer colors Pie and Time Series widgets in Builder can now inherit the colors of the layer they visualize. When a widget uses the same column your layer is styled by, it automatically picks up the layer's category colors, so the widget, the map, and the legend all match, with no extra setup. A category shown in purple on the map is purple in the widget too. Colors are now stable within a session as well: even without a layer match, each category keeps its color as you zoom, pan, and filter, instead of shifting with the data order. This applies to the Pie widget and to the Time Series widget with Split by. Learn more in the [Pie widget](/carto-user-manual/maps/widgets/pie-widget.md) and [Time Series widget](/carto-user-manual/maps/widgets/time-series-widget.md) documentation.
{% endupdate %} {% endupdates %} --- # 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.carto.com/whats-new/q3-2026.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.