mcp-brasil

Why we included this project

Teams building agents that answer questions about Brazil face a familiar problem: the data they need sits scattered across dozens of separate government APIs, and wiring each one by hand is slow. mcp-brasil bundles seventy public Brazilian sources into a single Model Context Protocol server with more than five hundred tools across fifteen topic areas, so an agent like Claude or GPT can ask about legislation, court rulings, spending, or election results in natural language. Most of the sources need no API key at all, and the heavy ones, such as the TSE election data and the INEP school census, are cached locally in embedded DuckDB for fast SQL-style queries. A planning tool can combine several sources into one execution, which helps when an answer spans datasets, like checking a deputy's spending against their voting record.