MAPA (Agricultura) MCP for AI. Query official Brazilian agri-data and datasets.

Your AI client uses the search_packages tool to find specific dataset packages across Brazil's Ministry of Agriculture data instantly by supplying keywords or filters like an organization name. You can run list_packages for a basic inventory list, or you might use list_tags to see every descriptive tag used throughout the entire portal, helping you narrow down your search criteria from the start.

When you need to map out exactly what data is available and where it comes from, you first run list_groups, which spits back a list of all high-level categories for the records. If you want to track who published the records, you'll use list_organizations to get names and IDs of every contributing department.

You can then drill down using get_organization on any specific ID to pull detailed provenance information about that government source. The system also lets you check which data groups are involved by running list_groups, letting you categorize the scope before you even look at a package.

To understand what's inside a dataset container, you run get_package using a known package ID; this pulls all the metadata for an entire collection, telling you its full scope and update history. If you know which department published it, but don’t want to list every single group or organization, you can use search_packages with a filter that targets the source itself.

You'll find data packaged under various names; if you run list_packages, you get the basic IDs needed for further inspection.

If you know exactly what file type you need—say, a CSV or an XLS sheet—you don't have to guess. You use get_resource on an individual resource ID within a package to fetch its metadata and direct download link. This tool confirms the file format and gives you the exact means to pull down the data itself.

The ability to cross-reference everything is key: if you find a promising dataset through searching, you can immediately run get_package to validate its contents before using get_resource on any of the underlying files it contains.

You never have to guess what's available; running search_packages allows you to filter thousands of records by keyword or organization name. If your initial search is broad, you can always run list_organizations and then use get_organization on a specific result to validate the source department. Once you identify a package ID via list_packages, running get_package provides the complete metadata profile, letting you confirm things like data coverage area or last update date.

To nail down the actual file download, remember that after getting the package details, you must run get_resource to get the specific link and format information for that file.

This server connects your agent directly to the raw structure of Brazilian governmental data. You'll use list_groups to map high-level categories, then potentially list_organizations to see all contributing entities; you can always follow up with get_organization if you want deep details on one source. If a search term is vague, running search_packages helps narrow it down using keywords or filters like 'agrofit.' Once you've found the right dataset container, use list_packages to get its ID, then feed that into get_package for the full metadata report.

For every single file inside that package, you run get_resource to grab the download details and format confirmation. You can check what descriptive tags are applied everywhere by running list_tags, which provides a complete list of classification terms across all records.