BCB SGS MCP Server

Source-reviewed, not functionally tested

These details were checked against primary sources. Skill Road has not performed a practical client functionality test. Features and setup reflect project documentation, not independently confirmed operation.

Traceable research into Brazil's economic data

BCB SGS MCP Server is the independent bcb-br-mcp project maintained by Sidney da Silva Pereira Bissoli. It connects MCP-capable AI clients to public data services of Banco Central do Brasil, Brazil's central bank. The institution supplies the underlying data but does not operate or endorse this connector. Keep that distinction visible when assessing the project. It supports public research and analysis, not access to bank accounts, payment execution or trading operations. An assistant gains a route to evidence rather than authority to act on someone's finances.

Separate observed values from market expectations

The documentation covers SGS time series, the Focus market-expectations survey and PTAX exchange rates. These services answer different questions. An observed inflation figure is not a forecast, and a survey expectation is not a binding statement about what will happen next. Establish whether the task describes a data vintage, compares a historical period or investigates expectations. Ask the assistant to explain that choice before combining results in one table or turning several different measures into an apparently definitive economic conclusion.

Meaningful comparisons retain series identifiers, units, frequency and observation periods. For interest rates, the reporting convention can matter as much as the value. Exchange-rate comparisons need an explicit reference currency and quote type. Check whether a selected series represents a stock, a rate of change or an index. Calculating percentage change for a variable already expressed as a rate may answer a different question from measuring the development of a price index. Statistical tools are useful, but their output still deserves a short methodological plausibility check before it becomes a published claim.

Preserve provenance with the analysis

The project says successful responses include source and timing information. Retain that material alongside the analysis rather than saving only the assistant's narrative. Extraction time for cached reference material may differ from the time you asked the question. When several services contribute to an answer, their sources should remain distinguishable. This allows another reader to understand which parts follow directly from observations and which depend on the assistant's selection or calculation. A citation attached to the final paragraph is less useful if the underlying comparison cannot be reconstructed.

Understand setup, data flow and separate licensing

The maintainer documents local stdio operation and a hosted HTTP endpoint without an API key. The reviewed npm package requires Node.js 22 or newer; the older minimum in the README does not override that package requirement. A local process still contacts external BCB interfaces, while hosted access introduces the community operator into the request path. The project identifies MIT for its code and ODbL for data and the curated catalog. Read those conditions separately before distributing a derived database or publishing material based on retrieved records.

Suitable uses and responsible boundaries

The integration can support market observation, teaching and reproducible research about Brazil. It does not replace inspection of original data or individualized financial advice. Keep private account information and internal trading strategies out of requests sent to a public server. Begin with a small date range, inspect the source and expand the task only after validating the result. If the available evidence cannot support a conclusion, the assistant should say so rather than filling a missing observation or an ambiguous definition with a plausible estimate that conceals the original uncertainty.

Source review and verification limits

This entry was reviewed against the linked primary sources on 2 October 2026. Features and setup instructions reflect the project's documentation rather than a complete functional test performed by Skill Road. Before production use, connect with non-sensitive example data, inspect the enabled tools, and compare responses with their original source. Changes to the service or your client may require a fresh review. The review date is not an uptime guarantee.

An MCP client. The reviewed npm package requires Node.js 22 or newer and npm locally, overriding the older README requirement. Network access to public BCB interfaces.

Use a local stdio entry with command npx and args ["-y", "bcb-br-mcp@1.15.1"]. Alternatively connect https://bcb.sidneybissoli.com/mcp as a remote HTTP endpoint. Begin with a narrowly scoped public time series.

npx -y bcb-br-mcp@1.15.1

The project documents public queries without an account, login or API key. A self-hosted HTTP endpoint still needs appropriate access control.

Read public SGS time series, Focus expectations, PTAX rates and metadata, and request statistical analysis. No documented account access, payments or trading operations.

Local queries send parameters directly to public BCB APIs; remote use additionally passes through the community host. The project says tool arguments are not logged; the hosting provider may process connection metadata.

An independent project, not endorsed by BCB. Distinguish observations, expectations and rate types. Preserve provenance, units, frequency and extraction time; do not send confidential financial data.