Ingest an AI / LLM call trace

Records one LLM call: model, token counts, cost, latency, status. project_id is derived from the API key.

Authorization

ApiKeyAuth

Authorization<token>

Format: "Bearer kw_"

In: header

Request Body

application/json

AI trace payload

TypeScript Definitions

Use the request body type in TypeScript.

agent_run_id?string

cost_usd?number

Range0 <= value

error_message?string

Lengthlength <= 2000

input_tokens?integer

Range0 <= value

latency_ms?integer

Range0 <= value

metadata?

model*string

Length1 <= length <= 200

output_tokens?integer

Range0 <= value

provider?string

Lengthlength <= 50

request_id?string

Lengthlength <= 200

status*string

Value in

"success"
"error"
"timeout"

timestamp?string

total_tokens?integer

Range0 <= value

Response Body

`application/json`

curl -X POST "https://example.com/ingest/ai" \  -H "Content-Type: application/json" \  -d '{    "model": "string",    "status": "success"  }'

{  "id": "string",  "timestamp": "string"}

Ingest an AI / LLM call trace

Authorization

Request Body

Response Body

201application/json

400application/json

401application/json

500application/json

`application/json`

`application/json`

`application/json`

`application/json`