List an Organization's AI Conversations

ExperimentalThis API is under active development and may change.
GET /api/0/organizations/{organization_id_or_slug}/agents/conversations/

Return AI conversations ordered by latest span time.

query uses Sentry search syntax.

Span filters

  • Description: "payment failed"; status: span.status:[error,internal_error].
  • Operation: gen_ai.operation.type:tool; tool: gen_ai.tool.name:get_weather.
  • Duration: span.duration:>2s; ID: gen_ai.conversation.id:"session:123" or conversation.conversationId:"session:123".

Conversation filters

  • Calls: conversation.llmCalls or conversation.messages, and conversation.toolCalls; failures: conversation.errors and conversation.toolErrors.
  • Usage: conversation.inputTokens, conversation.outputTokens, conversation.totalTokens, and conversation.totalCost.
  • Timespan: conversation.timeSpan measures elapsed time between first and last AI spans; conversation.generationDuration sums LLM call durations. Both use milliseconds.

Use numeric comparisons such as conversation.toolCalls:>2. Queries return conversations, not spans. Each AND condition may match a different span. Returned totals include all AI spans in selected project, environment, and time filters, not only matching spans.

Negation means no span matches, including when an attribute is absent. Payloads are searchable only when recorded and not scrubbed. Payload failure text does not mark a span as failed; set span.status:error for reliable failure search.

Path Parameters

organization_id_or_slug (string)
REQUIRED

The ID or slug of the organization the resource belongs to.

Query Parameters:

project (array(undefined))

The IDs or slugs of projects to filter by. Project slugs are unique within each organization. Omit this parameter to include all accessible projects. -1 is also accepted to include all accessible projects. For example, the following are valid parameters:

  • /?project=1234&project=56789
  • /?project=android&project=javascript-react
  • /?project=-1
environment (array(string))

The name of environments to filter by.

statsPeriod (string)

The period of time for the query, will override the start & end parameters, a number followed by one of:

  • d for days
  • h for hours
  • m for minutes
  • s for seconds
  • w for weeks

For example, 24h, to mean query data starting from 24 hours ago to now.

start (string)

The start of the period of time for the query, expected in ISO-8601 format. For example, 2001-12-14T12:34:56.7890.

end (string)

The end of the period of time for the query, expected in ISO-8601 format. For example, 2001-12-14T12:34:56.7890.

query (string)

Sentry search syntax matched against spans. A conversation is returned when any span in it matches. Summary fields include all spans in selected projects and time range, not only matching spans.

cursor (string)

A pointer to the last object fetched and its sort order; used to retrieve the next or previous results.

per_page (integer)

Number of conversations to return per page. Defaults to 10; maximum is 100.

Scopes

<auth_token> requires one of the following scopes:
  • org:admin
  • org:read
  • org:write
Copied
curl https://sentry.io/api/0/organizations/{organization_id_or_slug}/agents/conversations/ \
 -H 'Authorization: Bearer <auth_token>'
RESPONSESCHEMA
Copied
[
  {
    "conversationId": "01JQZ4W8X7J2Q9B4R5M6N7P8T9",
    "title": "Check San Francisco weather",
    "projectId": 1,
    "projects": [
      {
        "id": 1,
        "name": "Weather Assistant",
        "slug": "weather-assistant"
      }
    ],
    "webUrl": "https://sentry.io/organizations/org-slug/explore/agents/conversations/01JQZ4W8X7J2Q9B4R5M6N7P8T9/?project=1",
    "flow": [
      "Weather Assistant"
    ],
    "errors": 0,
    "llmCalls": 2,
    "toolCalls": 1,
    "totalTokens": 485,
    "inputTokens": 320,
    "outputTokens": 165,
    "totalCost": 0.0042,
    "timeSpan": 2500,
    "generationDuration": 1250.5,
    "startTimestamp": 1743465600000,
    "endTimestamp": 1743465602500,
    "traceCount": 1,
    "traceIds": [
      "4c79f60c11214eb38604f4ae0781bfb2"
    ],
    "firstInput": "What is the weather in San Francisco?",
    "lastOutput": "It is currently 18°C and sunny in San Francisco.",
    "user": {
      "id": "123",
      "email": "user@example.com",
      "username": "example-user",
      "ip_address": "192.0.2.1"
    },
    "toolNames": [
      "get_weather"
    ],
    "toolErrors": 0
  }
]
Was this helpful?