The drill-down behind one record: exact prompt/response/reasoning (model) or inputs/outputs (tool)

GET

https://api.recursiv.io/api/v1/evidence/records/:id

idstringRequiredformat: "uuid"

target_typeenumRequired

Allowed values:

target_idstringRequiredformat: "uuid"

kindenumOptionalDefaults to model_call

Allowed values:

The captured detail for the record

401

Unauthorized Error

404

Not Found Error

$	curl -G https://api.recursiv.io/api/v1/evidence/records/id \
>	-H "Authorization: Bearer <token>" \
>	-H "Content-Type: application/json" \
>	-d target_type=agent \
>	-d target_id=7c9e6679-7425-40de-944b-e07fc1f90ae7 \
>	-d kind=model_call

1	{
2	"record_id": "3fa85f64-5717-4562-b3fc-2c963f66afa6",
3	"target_type": "agent",
4	"target_id": "7c9e6679-7425-40de-944b-e07fc1f90ae7",
5	"kind": "model_call",
6	"timestamp": "2024-06-10T15:45:30Z",
7	"prompt": "Translate the following English text to French: 'Hello, how are you?'",
8	"response": "Bonjour, comment ça va ?",
9	"reasoning": "Used a standard French translation model to convert the greeting.",
10	"model": "gpt-4-translation-v1"
11	}