ReferenceAdmin
Individual moderation decisions behind the counts by category (platform admins and auditors)
GET
/v1/admin/analytics/moderation-eventsThe drill-down of analytics moderation: one entry per moderated stage of an answer whose decision was not pass, newest first, with the category that triggered, the score and whether an uncalibrated provider's block only flagged (downgraded). Content-free: no message text and no user identity (ADR-0010). Test chats are excluded.
Query Parameters
from?string
First UTC day (default 29 days before to)
Format
dateto?string
Last UTC day, inclusive (default today); at most 366 days after from
Format
datedecision?string
Value in
- "block"
- "flag"
- "support"
- "error"
category?string
A category that triggered (the top category or any triggered one)
Value in
- "violence"
- "self_harm"
- "sexual"
- "sexual_minors"
- "harassment_hate"
- "illicit"
- "personal_data"
- "prompt_injection"
agentId?string
Format
uuidcursor?string
nextCursor from the previous page
limit?integer
Range
1 <= value <= 200Default
50Response Body
application/json
application/json
application/json
curl -X GET "https://example.com/v1/admin/analytics/moderation-events"{ "data": { "items": [ { "id": 0, "at": "2019-08-24T14:15:22Z", "agentId": "bc309ecf-5f66-4057-93c5-6611cc9cb7b2", "agentName": "string", "agentSlug": "string", "teamSlug": "string", "agentDeleted": true, "channel": "ui", "audience": "team", "stage": "input", "decision": "block", "topCategory": "string", "score": 0.1, "categories": [ "string" ], "downgraded": [ "string" ], "provider": "string", "calibrated": true, "latencyMs": 0 } ], "nextCursor": "string" }}