Quickstart

1. Log a completion

After any LLM call, POST exactly what you sent and got back. The free tier stores 100 captures.

curl https://parcela.dev/api/v1/completions \
  -H "authorization: Bearer $PARCELA_KEY" \
  -H "content-type: application/json" \
  -d '{
    "name": "checkout-classifier",
    "provider": "openai",
    "endpoint": "https://api.openai.com/v1/chat/completions",
    "request": { "model": "gpt-4o-mini", "messages": [ ... ] },
    "response": { "choices": [ ... ] }
  }'

2. Replay it exactly

Re-run a stored call byte-for-byte, or pass overrides to change the model or prompt and diff the output. Your provider key is used once and never stored.

curl https://parcela.dev/api/v1/replay \
  -H "authorization: Bearer $PARCELA_KEY" \
  -H "content-type: application/json" \
  -d '{
    "id": "<capture-id>",
    "provider_key": "<your-openai-key>",
    "overrides": { "model": "gpt-4o" }
  }'

3. List your captures

curl https://parcela.dev/api/v1/completions \
  -H "authorization: Bearer $PARCELA_KEY"