Brazil CNPJ → Company Markdown DD Card

Turn a list of Brazilian CNPJs into clean company due-diligence cards — Markdown for humans and LLM/RAG pipelines, structured JSON for your ERP, KYB or vendor-master cleanup.

• Keyless by default — uses the public BrasilAPI CNPJ JSON (Receita Federal open data), with automatic fallback to CNPJ.ws and ReceitaWS public endpoints
• Bring your own key for production volume — ReceitaWS token or CNPJ.ws commercial token
• Markdown card per company: Identity · Status · CNAE · Address · QSA (partners) · Source attribution
• RAG chunks (CHUNKS.jsonl) — one self-describing chunk per card section, ready to embed
• No contact harvesting — email, telephone, fax and DDD fields are stripped from every provider payload. This is a due-diligence tool, not a lead list.
• 256 MB default memory, no browser, official/public JSON APIs only
• Failed rows are free — invalid CNPJ, not found, rate-limited, provider errors

Quick start

{
  "cnpjs": ["33.000.167/0001-01", "00.000.000/0001-91"]
}

That is all you need — provider: auto tries BrasilAPI first and falls back automatically.

Input

Field	Type	Default	Notes
cnpjs	string[]	—	Punctuation OK. Digits-only values shorter than 14 are left-padded with zeros (spreadsheets drop them). Check digits are validated; invalid rows are free. Supports the 2026 alphanumeric CNPJ format.
datasetId	string	—	Read CNPJs from another run's dataset (cnpjField, else cnpj / CNPJ / taxId / documento)
keyValueStoreId + keyValueRecordKey	string	INPUT_CNPJS	JSON array, {"cnpjs": [...]} or one CNPJ per line
cnpjField	string	—	Field name in dataset items / JSON objects
provider	enum	auto	auto · brasilapi · cnpjws · receitaws
apiKey	secret	—	Optional ReceitaWS token (Authorization: Bearer) or CNPJ.ws commercial token (x_api_token)
keyProvider	enum	receitaws	Which provider your key belongs to in auto mode
stopOnAuthError	boolean	true	After a 401/403 from the keyed provider, skip it for the remaining CNPJs
includeQsa	boolean	true	Partners/administrators: name, role, entry date only
includeRagChunks	boolean	true	Write CHUNKS.jsonl
maxItems	integer	100	0 = no limit
maxConcurrency	integer	2	1–5
maxRetries	integer	3	429 (honours Retry-After), 5xx, timeouts
requestTimeoutSecs	integer	30	Per HTTP call

Provider order (auto)

1. BrasilAPI https://brasilapi.com.br/api/cnpj/v1/{cnpj} (keyless)
2. With apiKey: your keyed provider (ReceitaWS or CNPJ.ws commercial). Without: CNPJ.ws public
3. The remaining keyless endpoint (ReceitaWS or CNPJ.ws public)

Keyless CNPJ.ws / ReceitaWS endpoints are rate-limited (about 3 requests/minute), which is why they are fallbacks only.

Output

Dataset — one row per CNPJ:

cnpj, cnpjFormatted, razaoSocial, nomeFantasia, matrizFilial, situacao, situacaoDate, situacaoMotivo, situacaoEspecial, dataInicioAtividade, naturezaJuridica {code, description}, porte, capitalSocial, cnaePrincipal {code, codeFormatted, description}, cnaesSecundarios[], address {tipoLogradouro, logradouro, numero, complemento, bairro, municipio, uf, cep, codigoMunicipioIbge}, addressLine, simples, mei, qsa[] {name, role, since}, qsaCount, markdown, source, sourceUrl, sourceUpdatedAt, triedSources, status, errorClass, errorMessage, hint.

Views: Overview and Failed (not charged).

Key-value store:

• REPORT.md — all cards + a failed table
• CHUNKS.jsonl — {id, cnpj, razaoSocial, section, source, text, chunkIndex, chunkCount} per line
• SUMMARY / OUTPUT — counts by result class and source, charged events, duration, peak memory

Example card (excerpt)

## PETROLEO BRASILEIRO S A PETROBRAS

**CNPJ** `33.000.167/0001-01` · **ATIVA** · source: BrasilAPI

### CNAE (economic activity)

- **Principal:** `0600-0/01` — Extração de petróleo e gás natural
- **Secundários (5):** …

### Address

AVENIDA REPUBLICA DO CHILE, 65 — CENTRO — RIO DE JANEIRO/RJ — CEP 20031-170

errorClass values: invalid_input, not_found, rate_limited, unauthorized, forbidden, server_error, timeout, network, bad_response, skipped_auth.

Privacy

• Email, telephone, fax and DDD fields are dropped from every provider payload before processing, and every output string is scrubbed for email addresses.
• QSA contains names and roles only (plus entry date when the source has it) — no masked CPF fragments, no age bands, no contact data.
• Your apiKey is stored encrypted by Apify and is never logged or written to output.

Pricing (pay per event)

Event	Price	When
Actor start	$0.001	once per run
cnpj-card	$0.003	per successful company card

Invalid CNPJs, not found, rate-limited after retries, auth errors and provider failures are free.

Data sources

Company data originates from the Receita Federal do Brasil CNPJ open-data release, served by BrasilAPI, CNPJ.ws and ReceitaWS. Each card names the provider and endpoint it came from. Use of a keyed provider follows that provider's own terms.