List requests

Lists employee requests handled by the Harmony AI agent

List requests

get

Paginated list of requests: conversations the AI agent handled, each linked to the ticket it opened (if any) via ticket_id. resolution says who the request ended with: Harmony or a human.

Authorizations

AuthorizationstringRequired

Harmony access key, sent as Authorization: AccessKey <your-key>. Create keys in Settings > Access Keys. The tenant is resolved from the key.

Query parameters

pageinteger · min: 1OptionalDefault: 1

page_sizeinteger · min: 1 · max: 200OptionalDefault: 25

sort_bystring · enumOptionalDefault: queried_atPossible values:

sort_directionstring · enumOptionalDefault: descPossible values:

date_fromstring · date-time · nullableOptional

date_tostring · date-time · nullableOptional

resolutionstring[] · nullableOptional

Who the request ended with. resolved_by_harmony: Harmony resolved it (including tickets closed while assigned to Harmony). escalated: handed off to a human, whether the human resolved, closed or canceled it. pending_harmony, pending_human, pending: still open. Null: not yet determined, or withdrawn.

desk_idstring[] · nullableOptional

subdesk_idstring[] · max: 100 · nullableOptional

Team (subdesk) the conversation was attributed to, matched against the request's denormalized subdesk_id. The dashboards call this dimension team; the value is a service_desk subdesk id, the same id space the Team dashboard filter uses.

categorystring[] · nullableOptional

employee_idstring · nullableOptional

searchstring · min: 1 · max: 255 · nullableOptional

include_classificationbooleanOptionalDefault: false

classificationstring[] · nullableOptional

relevancestring[] · nullableOptional

ai_contributionstring[] · nullableOptional

human_contributionstring[] · nullableOptional

resolution_feedbackstring[] · nullableOptional

outcomestring[] · nullableOptional

How the AI contributed to the request (e.g. ai_resolved_direct, ai_assisted_submitted, manual_ticketed). Describes AI involvement, not who closed the request; use resolution for that.

handlingstring[] · nullableOptional

sourcestring[] · max: 20 · nullableOptional

Channel the conversation came in on, matched case-insensitively against the request's source. The dashboards call this dimension channel. Every request has a source, so unlike department an empty string here matches nothing rather than meaning 'unknown'.

departmentstring[] · max: 500 · nullableOptional

Department of the requesting employee, read from the employee record rather than the request. Lower-cased on both sides, and an empty string matches requests whose reporter has no department or is unknown - the same bucketing the Requests by Department metric uses.

deflected_kb_article_idstring[] · nullableOptional

referenced_kb_article_idstring[] · nullableOptional

Responses

200

Successful Response

application/json

Paginated list of requests.

total_countintegerRequired

403

Access denied

422

Validation Error

application/json

get/api/v1/conversations/requests

GET /agents/api/v1/conversations/requests HTTP/1.1
Host: api.harmony.io
Authorization: YOUR_API_KEY
Accept: */*

{
  "items": [
    {
      "id": "123e4567-e89b-12d3-a456-426614174000",
      "thread_id": "123e4567-e89b-12d3-a456-426614174000",
      "employee_id": "text",
      "desk_id": "text",
      "agent_name": "text",
      "category": "text",
      "subcategory": "text",
      "source": "text",
      "resolution": "text",
      "summary": "text",
      "ticket_id": "text",
      "ticket_type": "text",
      "entities": [
        "text"
      ],
      "symptoms": [
        "text"
      ],
      "queried_at": "2026-01-01T00:00:00.000Z",
      "resolution_at": "2026-01-01T00:00:00.000Z",
      "last_message_at": "2026-01-01T00:00:00.000Z",
      "created_at": "2026-01-01T00:00:00.000Z",
      "classification": "text",
      "classification_reasoning": "text",
      "relevance": "text",
      "ai_contribution": "text",
      "human_contribution": "text",
      "resolution_feedback": "text",
      "classified_at": "2026-01-01T00:00:00.000Z",
      "outcome": "text",
      "handling": "text",
      "search_match": {
        "location": "message",
        "snippet": "text",
        "field_label": "text"
      }
    }
  ],
  "total_count": 1
}