PNAD

Use this skill whenever the task needs to answer a question about Brazil with local official microdata — PNADC trimestral, PNADC anual visita 5, Censo 2022, TSE eleitorado, IPCA, or salário mínimo — using the `brasil` / `pnad` CLI. Trigger for any state-level income analysis, household income-band distribution, annual income-source composition (labor vs benefits vs pensions vs capital), dashboard snapshot, SQLite query, or when a journalist / analyst / LLM agent needs a read-mostly, statistically auditable answer grounded in the repo's local outputs instead of writing ad-hoc pandas code.

Use this skill for ad-hoc read-only SQL on the PNADC SQLite database produced by this repo (`data/outputs/brasil.sqlite`). Trigger when the user asks for a precise custom aggregate — a specific AVG/SUM/COUNT, a specific UF or demographic cut, an arbitrary filter — that `brasil dashboard` or `brasil renda-por-faixa-sm` cannot express. Provides a PT/EN column dictionary, ten ready-made safe queries, and LLM guardrails. Companion to `brasil-cli-analyst` for the "exact number for X" use case.