Sign in
scout

Get Scout Api Activity

Changed on

Return cursor-paginated API activity for the calling customer.

Each row represents one API-originated Scout agent run. task_id is the identifier returned by POST /scout/tasks and can be used to filter to one task. status is aggregated from the run's LLM calls: error if any call is in error, otherwise success if any call succeeded, else pending. Token counts, cost and status may be updated after the run first appears, so usage should be treated as subject to late updates rather than final.

Use started_after (inclusive), started_before (exclusive), status, and task_id to filter the activity. Follow cursor pagination by walking page_info.next until page_info.has_next is false. Pagination is keyset-ordered by start time over data that can change while it is being read; for reconciliation, re-read overlapping time windows and deduplicate by task_id. Cost is calculated as input and output tokens multiplied by the configured rates.

get/scout/api-usage

Request

  • Base URL: https://api.harmonic.ai (the document declares no server; this is the origin it was published from)
  • URL: https://api.harmonic.ai/scout/api-usage
  • Auth: none declared

Query parameters

cursorstring nullable

Pagination cursor

page_sizeinteger

Page size

started_afterstring date-time nullable

Include runs started at or after this datetime

started_beforestring date-time nullable

Include runs started before this datetime

status'pending' | 'success' | 'error'

Lifecycle status of a single LLM call.

These are LangSmith's run-status strings, which both writers pass through verbatim: agent-graph's callback and the LangSmith sync. The scout_api_run_aggregates view rolls rows up on the same two terminal literals, so keep the three in step.

Filter by run status

task_idstring nullable

Filter to the task ID returned by POST /scout/tasks

Response

Successful Response

Changes