Skip to main content
POST
Create Decision Scorer Version

Authorizations

Splunk-AO-API-Key
string
header
required

Path Parameters

scorer_id
string<uuid4>
required

Body

application/json

The question and model for a new decision scorer version.

The inherited scoreable_node_types, output_type and input_type are optional, but each must match the scorer when sent. The service checks these fields, and it checks answer_options against the stored output_type.

model_name
string
required
Required string length: 1 - 255
question
string
required
Required string length: 1 - 2000
num_judges
integer | null
scoreable_node_types
string[] | null
cot_enabled
boolean | null
output_type
enum<string> | null

Enumeration of output types.

Available options:
boolean,
categorical,
count,
discrete,
freeform,
percentage,
multilabel,
retrieved_chunk_list_boolean,
boolean_multilabel,
boolean_probability
input_type
enum<string> | null

Enumeration of input types.

Available options:
basic,
llm_spans,
retriever_spans,
sessions_normalized,
sessions_trace_io_only,
tool_spans,
trace_input_only,
trace_io_only,
trace_normalized,
trace_output_only,
agent_spans,
workflow_spans
answer_options

Response

Successful Response

id
string<uuid4>
required
version
integer
required
scorer_id
string<uuid4>
required
created_at
string<date-time>
required
updated_at
string<date-time>
required
generated_scorer
GeneratedScorerResponse · object | null
registered_scorer
CreateUpdateRegisteredScorerResponse · object | null
finetuned_scorer
FineTunedScorerResponse · object | null
luna3_scorer
Luna3ScorerResponse · object | null
decision_scorer
DecisionScorerDB · object | null

A row of the decision_scorers table.

answer_options depends on the output type of the scorer:

  • categorical: a list of labels (list[str]).
  • discrete: a list of DecisionLevel.
  • boolean_probability: None.
model_name
string | null
num_judges
integer | null
cot_enabled
boolean | null
runtime
string | null

The runtime a code metric version runs on, such as python3.14. Null for a version saved before runtimes were recorded, which runs on the legacy Python 3.12 runtime.

chain_poll_template
ChainPollTemplate · object | null

Template for a chainpoll metric prompt, containing all the info necessary to send a chainpoll prompt.

scoreable_node_types
string[] | null
output_type
enum<string> | null

Enumeration of output types.

Available options:
boolean,
categorical,
count,
discrete,
freeform,
percentage,
multilabel,
retrieved_chunk_list_boolean,
boolean_multilabel,
boolean_probability
allowed_model
boolean | null
created_by
string<uuid4> | null