Quick tests

Smoke-test the Sombra stack with ready-made curl examples.

Prerequisites

• Stack running: make up
• jq installed (output is piped through it for pretty printing)

Run the examples

make call-analyzer   # hit the Presidio analyzer directly (:5002)
make call-litellm    # hit the full proxy masking path (:4000), needs .env sourced
make test-call       # run both sequentially

make test-call is the fastest way to verify the whole chain in one shot.

Expected analyzer output

make call-analyzer detects entities in the sample text:

Entity	Confidence
SOMBRA_CPF	0.95 (1.00 with context)
SOMBRA_CEP	0.95
SOMBRA_CNPJ	0.95
SOMBRA_ADDRESS	0.75
PERSON / LOCATION	0.85 (spaCy NER)

Expected masked prompt

make call-litellm sends a chat request through the full masking path. The LiteLLM logs show the prompt reaching the provider already masked:

"content": "Consult my order in the name of João, CPF <SOMBRA_CPF>, living in Curitiba, CEP <SOMBRA_CEP>."

Run them by hand

curl -s http://localhost:5002/analyze -H "Content-Type: application/json" \
  -d '{"text":"Meu CPF é 123.456.789-00, CEP 80000-000, CNPJ 12.345.678/0001-90, Rua das Flores, 123.","language":"pt"}' | jq .

curl -s http://localhost:4000/v1/chat/completions \
  -H "Authorization: Bearer $LITELLM_MASTER_KEY" \
  -H "Content-Type: application/json" \
  -d '{"model":"opencode-zen/deepseek-v4-flash","messages":[{"role":"user","content":"Consult my order in the name of João, CPF 123.456.789-00, living in Curitiba, CEP 80000-000."}]}' | jq .

Next

These are smoke tests. For the real test suite, see Testing.