Skip to main content
POST
Create New Evaluator Rule

Authorizations

Authorization
string
header
required

Bearer authentication header of the form Bearer <token>, where <token> is your auth token.

Body

application/json

Request body for creating a Rule evaluator with filters

name
string
required

Name of the Rule evaluator

application_id
string<uuid>
required

Application UUID

evaluator_id
string<uuid>
required

UUID of the corresponding evaluator

mapped_input_keys
object | null

Mapping of input field names to verbatim span attribute keys (framework-specific). Mutually exclusive with mapped_semantic_names — provide exactly one.

mapped_semantic_names
object | null

Mapping of input field names to semantic concept names (framework-agnostic). Same structure as mapped_input_keys but values are semantic concepts instead of literal attribute keys. The enrichment worker resolves each concept to all mapped attribute keys at evaluation time via the semantic mapping cache. Valid concepts: input, output, system_instructions, retrieval_context, tool_input, tool_output. Mutually exclusive with mapped_input_keys — provide exactly one.

filters
object

Filters for applying Rule evaluator

backfill
boolean
default:false

Whether to backfill already ingested spans

sampling_rate
number<float>
default:1

Fraction of matching spans on which this evaluator runs, in the half-open range (0.0, 1.0]. Defaults to 1.0 (every matching span). Traces are always persisted even when downsampling is enabled. To pause evaluation entirely, create the rule and then PATCH it with enabled=false — a small sampling_rate is not a pause mechanism (and enabled is not accepted on this create request).

Required range: 0 < x <= 1
backfill_start_time
string<date-time> | null

Starting point for backfill in case backfill is not required for the entire period

Response

Details of the created evaluator rule

Response object for standard API responses.

api_version
enum<string>
default:3.0

API version.

Available options:
2.0,
3.0
kind
enum<string>
default:NORMAL

Type of response, indicating a normal response.

Available options:
NORMAL
data
object

Rule Evaluator