Brazilian Central Bank (BCB) MCP

What Brazilian Central Bank (BCB) MCP MCP server does

The Brazilian Central Bank (BCB) MCP MCP server gives AI assistants access to public economic and financial data from Banco Central do Brasil. Its main data source is SGS, which provides time series for indicators such as Selic, IPCA, exchange rates, GDP, employment, credit, fiscal measures, and monetary aggregates. The server also exposes the Focus survey for market expectations and PTAX official exchange-rate quotations.

The package includes 15 MCP tools, three read-only resources, and three prompts. Responses include provenance metadata, and structured results are available alongside JSON text output where the tool defines an output schema. The catalog contains more than 150 curated, source-verified series, organized by economic category.

How it works

Agents can first search the local catalog by keyword or list popular series by category, then use the returned SGS code to retrieve data. Historical queries accept date ranges, while another tool returns the latest N observations. Metadata queries provide a series name, unit, frequency, source, category, and direct API URLs without retrieving the full value history.

Analysis tools calculate percentage changes and descriptive statistics for one series or rank two to five series over the same period. The current-indicators shortcut fetches recent Selic, IPCA, USD/BRL, and IBC-Br values independently, so one failed indicator does not invalidate the others. Additional tools cover Focus expectations by indicator or Copom meeting, available Focus references, PTAX quotations, and currencies published by the BCB.

The server handles transient SGS failures with up to three attempts and exponential backoff. Long date windows and upstream observation limits are handled by slicing requests and merging results where applicable. Series can also be harmonized to coarser monthly, quarterly, or annual frequencies using the documented conventions.

Setup and configuration

The Brazilian Central Bank (BCB) MCP MCP server can run locally with the npm package and stdio transport:

{
  "mcpServers": {
    "bcb-br": {
      "command": "npx",
      "args": ["-y", "bcb-br-mcp"]
    }
  }
}

It is explicitly documented for Claude Desktop, with configuration paths for Windows and macOS. A global npm installation is also supported through npm install -g bcb-br-mcp, followed by the bcb-br-mcp command. Alternatively, MCP clients can connect to https://bcb.sidneybissoli.com/mcp without local installation. No API key, credentials, or signup are required for the documented public-data access.

Tools and capabilities

• Discover curated BCB series by category or accent-insensitive keyword.
• Retrieve complete or date-bounded series histories and recent observations.
• Inspect series metadata and access direct BCB API URLs.
• Fetch a compact set of current Brazilian economic indicators.
• Calculate single-series variation, statistics, and compounded rates where applicable.
• Compare multiple series over a common period.
• Query Focus expectations and PTAX currency quotations.
• Attach catalog resources for popular series, categories, and principal indicators.
• Use prompts for current indicators, an economic panorama, or inflation comparisons.

Limitations and notes

Catalog search is limited to the curated collection rather than the full SGS catalog, which contains many more series. A series code may therefore need to be found through the BCB SGS portal when the local catalog has no match. Date and value formatting follows the server output: dates use dd/MM/yyyy, and numeric values use a decimal point.

The public BCB endpoints are used without authentication, and no request limit is stated; usage is best-effort. A persistent upstream failure produces an MCP error, while missing data or an HTTP 404 is reported as an unavailable series or period. Comparison requires two to five series and a shared date range, and variation requires at least two observations. Focus and PTAX capabilities depend on the data published by their respective BCB sources.