Loading ToolBench
MCP server for the Brazilian Central Bank (Banco Central do Brasil, BCB): SGS time series (Selic, IPCA, exchange rates, GDP and a curated catalogue of indicators), Focus market expectations, PTAX exchange rates and the Olinda open-data platform, with a provenance block on every response.
BCB BR MCP presents a well-curated domain-specific tool collection for Brazilian Central Bank data. Tool naming follows verb_noun conventions (sgs_series, buscar_series, obter_catalogo, etc.), which is strong. Descriptions are present and substantive (averaging ~100-150 chars per tool), clearly contextualizing what each tool does. Input schemas are consistently defined with parameter types and descriptions. However, there are gaps: output schemas are not explicitly documented in the tool definitions, error handling guidance is minimal, and some parameter descriptions lack constraint details (e.g., date format strings for 'data' parameters, numeric bounds for array lengths). The tools compose well for financial time-series analysis (search → retrieve → transform → analyze pipeline), though some multi-step operations could be consolidated. No security issues detected (all tools are READ_ONLY). The catalog verification note in tools.ts demonstrates serious attention to data correctness and provenance, which is commendable. Per-tool variance is moderate: naming and schema are consistent; descriptions vary slightly but remain adequate.
Alinha múltiplas séries por datas comuns
Busca séries do catálogo curado por nome ou palavras-chave
Calcula correlação entre duas ou mais séries temporais
Deflaciona uma série temporal usando um deflator (ex: IPCA)
Calcula estatísticas descritivas de uma série temporal
Retorna as 5 expectativas de mercado mais recentes para um indicador Focus
Harmoniza múltiplas séries para a mesma frequência/periodicidade
Retorna o catálogo completo de séries populares do BCB
Output schemas not explicitly documented in tool definitions. Clients cannot see what fields each tool returns, requiring either API exploration or internal code review to understand response structure.
Date parameter descriptions mention format (e.g., 'MM-DD-YYYY' for ptax_diaria, 'YYYY-MM-DD' for others) but lack explicit validation error guidance. LLMs may pass malformed dates; error responses should state the exact format required and show an example.
Inferred effective spec: 2026-07-28+.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | D | 54 | 2026-07-28+ | v2 |
Retorna a taxa de câmbio PTAX do dia especificado
Busca valores de uma série temporal do SGS (Sistema de Gerenciamento de Séries Temporais) do Banco Central
Retorna os últimos N valores de uma série
Calcula a variação acumulada de uma série em período especificado
Numeric array parameters (e.g., 'series' in correlacao_series, harmonizar_series, alinhar_series) lack explicit bounds documentation. Descriptions do not state minimum/maximum array length or maximum total series count. Large requests could cause timeouts or memory exhaustion.
Enum constraints for string parameters (e.g., 'metodo' in correlacao_series, variacao_acumulada; 'frequencia' and 'deflator' in other tools) are mentioned in descriptions as examples ('ex: pearson, spearman') but not declared as formal JSON Schema enums. LLMs may hallucinate unsupported values.
Error handling and recovery guidance minimal. Tool descriptions do not explain what happens on invalid input (e.g., non-existent serie codigo, malformed date, out-of-range quantidade) or how to retry/correct.
No indication of which parameters are optional vs. required. 'dataInicio' and 'dataFim' marked as optional in descriptions but unclear if their omission changes tool behavior (e.g., returns full range vs. defaults to recent window).