GET
/
metrics
curl --request GET \
  --url https://api.honeyhive.ai/metrics \
  --header 'Authorization: Bearer <token>'
[
  {
    "name": "<string>",
    "criteria": "<string>",
    "code_snippet": "<string>",
    "prompt": "<string>",
    "task": "<string>",
    "type": "custom",
    "description": "<string>",
    "enabled_in_prod": true,
    "needs_ground_truth": true,
    "return_type": "boolean",
    "threshold": {
      "min": 123,
      "max": 123
    },
    "pass_when": true,
    "_id": "<string>",
    "event_name": "<string>",
    "event_type": "<string>",
    "child_metrics": [
      {}
    ]
  }
]

Authorizations

Authorization
string
header
required

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

Query Parameters

project_name
string
required

Project name associated with metrics

Response

200 - application/json
description
string
required

Short description of what the metric does

name
string
required

Name of the metric

return_type
enum<string>
required

The data type of the metric value - "boolean", "float", "string"

Available options:
boolean,
float,
string
task
string
required

Name of the project associated with metric

type
enum<string>
required

Type of the metric - "custom", "model", "human" or "composite"

Available options:
custom,
model,
human,
composite
_id
string

Unique idenitifier

child_metrics
object[]

Child metrics added under composite events

code_snippet
string

Associated code block for the metric

criteria
string

Criteria for human or composite metrics

enabled_in_prod
boolean

Whether to compute on all production events automatically

event_name
string

Name of event that the metric is set to be computed on

event_type
string

Type of event that the metric is set to be computed on

needs_ground_truth
boolean

Whether a ground truth (on metadata) is required to compute it

pass_when
boolean

Threshold for boolean metrics to decide passing or failing in tests

prompt
string

Evaluator prompt for the metric

threshold
object

Threshold for numeric metrics to decide passing or failing in tests