> ## Documentation Index > Fetch the complete documentation index at: https://docs.labelbox.com/llms.txt > Use this file to discover all available pages before exploring further. # Read pre-aggregated session, usage, and environment analytics > Returns totals, a time series, and default breakdowns for one metric family over a time window. Figures are kept up to date as work happens rather than recomputed from session events, so the cost of a request depends on the window rather than on how much history the organization has, and every response reports its own freshness through as_of and lag_seconds. Percentiles are approximate. scope defaults to the caller's own organization, which is the scope an organization member reads; tenant and global need a tenant-administrator identity and answer 403 otherwise. ## OpenAPI ````yaml /managed-agents/openapi.yaml get /managed-agents/v1/analytics/{family} openapi: 3.1.0 info: title: Managed Agents API version: 0.14.0 description: >- The Recursion Managed Agents REST API. Authenticate with a Recursion API key as a bearer token. An organization-scoped key acts in its own organization and needs nothing else; a tenant-scoped key must also send `x-organization-id` with an organization id or `default`. Field names follow each operation's published schema. servers: - url: https://api.recursion.labelbox.com security: - bearerAuth: [] paths: /managed-agents/v1/analytics/{family}: get: tags: - Analytics summary: Read pre-aggregated session, usage, and environment analytics description: >- Returns totals, a time series, and default breakdowns for one metric family over a time window. Figures are kept up to date as work happens rather than recomputed from session events, so the cost of a request depends on the window rather than on how much history the organization has, and every response reports its own freshness through as_of and lag_seconds. Percentiles are approximate. scope defaults to the caller's own organization, which is the scope an organization member reads; tenant and global need a tenant-administrator identity and answer 403 otherwise. operationId: managedAgentsGetAnalytics parameters: - description: >- Read locality. workspace is the caller's own organization; tenant aggregates every workspace in the parent tenant organization and requires a Tenant Admin role. in: query name: scope schema: default: workspace description: >- Read locality. workspace is the caller's own organization; tenant aggregates every workspace in the parent tenant organization and requires a Tenant Admin role. enum: - workspace - tenant type: string - description: >- With scope=tenant, narrow to one workspace of the tenant. Ignored at workspace scope, where the authenticated organization is authoritative. in: query name: workspace_id schema: description: >- With scope=tenant, narrow to one workspace of the tenant. Ignored at workspace scope, where the authenticated organization is authoritative. type: string - description: >- RFC 3339 start of the window. Defaults to seven days before to. Widened outward to a bucket boundary. in: query name: from schema: description: >- RFC 3339 start of the window. Defaults to seven days before to. Widened outward to a bucket boundary. type: string - description: RFC 3339 exclusive end of the window. Defaults to now. in: query name: to schema: description: RFC 3339 exclusive end of the window. Defaults to now. type: string - description: >- Series bucket width. Defaults to hourly for windows up to two days and daily beyond that. in: query name: granularity schema: description: >- Series bucket width. Defaults to hourly for windows up to two days and daily beyond that. enum: - hour - day type: string - description: >- Filter to work under sessions launched from this agent, including its subagents. in: query name: launched_agent_id schema: description: >- Filter to work under sessions launched from this agent, including its subagents. type: string - description: >- Filter to work performed by this agent, whether it was launched directly or delegated to. in: query name: executing_agent_id schema: description: >- Filter to work performed by this agent, whether it was launched directly or delegated to. type: string - description: Filter to one immutable agent version. in: query name: executing_agent_version_id schema: description: Filter to one immutable agent version. type: string - description: Filter to one provider model id. in: query name: model schema: description: Filter to one provider model id. type: string - description: >- Filter to one tool. Note that token and cost figures are model spend and are not attributable to a tool, so filtering by one reports only that tool's own direct cost. in: query name: tool_name schema: description: >- Filter to one tool. Note that token and cost figures are model spend and are not attributable to a tool, so filtering by one reports only that tool's own direct cost. type: string - description: >- Filter by outcome: tool call state for usage, startup result for environments, or immutable verdict for quality. in: query name: outcome schema: description: >- Filter by outcome: tool call state for usage, startup result for environments, or immutable verdict for quality. enum: - in_flight - completed - failed - ready - pending - pass - fail - not_applicable type: string - description: >- Filter environment analytics to one sandbox provider, e.g. runs or docker. self_hosted is accepted for historical queries. in: query name: sandbox_provider schema: description: >- Filter environment analytics to one sandbox provider, e.g. runs or docker. self_hosted is accepted for historical queries. type: string - description: Filter environment analytics to one environment. in: query name: environment_id schema: description: Filter environment analytics to one environment. type: string - description: >- Filter environment analytics by compute kind: cpu, or an accelerator type and name such as gpu:a100. in: query name: compute_class schema: description: >- Filter environment analytics by compute kind: cpu, or an accelerator type and name such as gpu:a100. type: string - description: >- Filter environment analytics to one machine type, or to the requested CPU and memory shape when the provider names none. in: query name: machine_type schema: description: >- Filter environment analytics to one machine type, or to the requested CPU and memory shape when the provider names none. type: string - description: Filter quality analytics to evaluations of one target agent. in: query name: target_agent_id schema: description: Filter quality analytics to evaluations of one target agent. type: string - description: Filter quality analytics to one immutable target-agent version. in: query name: target_agent_version_id schema: description: Filter quality analytics to one immutable target-agent version. type: string - description: >- Filter quality analytics to verdicts produced by one evaluator agent. in: query name: evaluator_agent_id schema: description: >- Filter quality analytics to verdicts produced by one evaluator agent. type: string - description: Filter quality analytics to one immutable evaluator-agent version. in: query name: evaluator_agent_version_id schema: description: Filter quality analytics to one immutable evaluator-agent version. type: string - description: Filter quality analytics to one stable rubric criterion key. in: query name: criterion_key schema: description: Filter quality analytics to one stable rubric criterion key. type: string - description: >- Filter quality analytics by the target task archetype frozen on the evaluation. in: query name: archetype schema: description: >- Filter quality analytics by the target task archetype frozen on the evaluation. type: string - description: >- Filter quality analytics by the target size bucket frozen on the evaluation. in: query name: size_bucket schema: description: >- Filter quality analytics by the target size bucket frozen on the evaluation. type: string - description: >- Return the series split into one line per value of this dimension, for a stacked chart, alongside the combined series. Capped at the top eight groups by the family's ranking measure. Omit for the combined series only. in: query name: series_group_by schema: description: >- Return the series split into one line per value of this dimension, for a stacked chart, alongside the combined series. Capped at the top eight groups by the family's ranking measure. Omit for the combined series only. type: string - description: Rows per breakdown. in: query name: limit schema: default: 25 description: Rows per breakdown. maximum: 100 type: integer - description: >- Metric family to read: managed_agent_usage for tool, token, and cost figures, managed_agent_turn for turn counts and turn duration percentiles, managed_agent_environment for sandbox startup latency, managed_agent_completion for first successful root-session completions, and managed_agent_quality for immutable evaluation and criterion verdicts. in: path name: family required: true schema: description: >- Metric family to read: managed_agent_usage for tool, token, and cost figures, managed_agent_turn for turn counts and turn duration percentiles, managed_agent_environment for sandbox startup latency, managed_agent_completion for first successful root-session completions, and managed_agent_quality for immutable evaluation and criterion verdicts. enum: - managed_agent_usage - managed_agent_turn - managed_agent_environment - managed_agent_completion - managed_agent_quality type: string responses: '200': content: application/json: schema: $ref: '#/components/schemas/ManagedAgentsAnalyticsResponse' description: >- Pre-aggregated analytics for one metric family, scope, and time window. Figures are kept up to date as work happens rather than recomputed from session events, so a response is fast regardless of how much history the organization has, and states its own freshness through as_of and lag_seconds. '400': description: The request was rejected by schema or semantic validation. content: application/json: schema: oneOf: - $ref: '#/components/schemas/ManagedAgentsApiErrorInvalidRequest' - $ref: '#/components/schemas/ManagedAgentsApiErrorInvalidScope' discriminator: propertyName: code mapping: invalid_request: $ref: '#/components/schemas/ManagedAgentsApiErrorInvalidRequest' invalid_scope: $ref: '#/components/schemas/ManagedAgentsApiErrorInvalidScope' '401': description: The caller is unauthenticated. content: application/json: schema: $ref: '#/components/schemas/ManagedAgentsApiErrorUnauthorized' '403': description: The caller lacks a required permission or capability. content: application/json: schema: $ref: '#/components/schemas/ManagedAgentsApiErrorForbidden' '429': headers: Retry-After: description: When to retry, as delay seconds or an HTTP date. schema: type: string description: A rate limit is exhausted. Honor `Retry-After`. content: application/json: schema: $ref: '#/components/schemas/ManagedAgentsApiErrorRateLimitExceeded' '500': description: An unexpected server-side failure. content: application/json: schema: oneOf: - $ref: '#/components/schemas/ManagedAgentsApiErrorInternalError' - $ref: >- #/components/schemas/ManagedAgentsApiErrorInvariantViolation discriminator: propertyName: code mapping: internal_error: $ref: '#/components/schemas/ManagedAgentsApiErrorInternalError' invariant_violation: $ref: >- #/components/schemas/ManagedAgentsApiErrorInvariantViolation '502': description: The service returned an invalid response. content: application/json: schema: $ref: '#/components/schemas/ManagedAgentsApiErrorBadGateway' '503': description: >- A dependency is unavailable. Consult details.retryable and Retry-After when present before retrying. content: application/json: schema: oneOf: - $ref: '#/components/schemas/ManagedAgentsApiErrorAnalytics' - $ref: >- #/components/schemas/ManagedAgentsApiErrorAnalyticsNotReady - $ref: >- #/components/schemas/ManagedAgentsApiErrorManagedAgentsUnavailable - $ref: >- #/components/schemas/ManagedAgentsApiErrorServiceUnavailable discriminator: propertyName: code mapping: analytics: $ref: '#/components/schemas/ManagedAgentsApiErrorAnalytics' analytics_not_ready: $ref: >- #/components/schemas/ManagedAgentsApiErrorAnalyticsNotReady managed_agents_unavailable: $ref: >- #/components/schemas/ManagedAgentsApiErrorManagedAgentsUnavailable service_unavailable: $ref: >- #/components/schemas/ManagedAgentsApiErrorServiceUnavailable '504': description: The service did not respond before the timeout. content: application/json: schema: $ref: '#/components/schemas/ManagedAgentsApiErrorGatewayTimeout' components: schemas: ManagedAgentsAnalyticsResponse: description: >- Pre-aggregated analytics for one metric family, scope, and time window. Figures are kept up to date as work happens rather than recomputed from session events, so a response is fast regardless of how much history the organization has, and states its own freshness through as_of and lag_seconds. properties: as_of: description: >- How current these figures are. Activity after this instant is counted in a later read. format: date-time type: string breakdowns: description: The family's default groupings, each capped by limit. items: $ref: '#/components/schemas/ManagedAgentsAnalyticsBreakdown' type: - array - 'null' family: description: Metric family these figures come from. type: string family_version: description: >- Version of the family's definition. A change here means how the figures are derived changed. format: int64 type: integer from: description: >- Inclusive start actually aggregated, widened outward to a bucket boundary. format: date-time type: string granularity: description: >- Width of each series bucket. Chosen from the window: hourly up to two days, daily beyond that. enum: - hour - day type: string lag_seconds: description: >- Seconds between as_of and now. A steady value under a minute is normal; a growing one means the figures are catching up. format: int64 type: integer scope: description: >- Read locality the figures were aggregated at, after authorization. This may be narrower than the scope requested. enum: - workspace - tenant - global type: string series: description: >- One point per time bucket, ascending. Buckets with no activity are omitted rather than zero-filled. items: $ref: '#/components/schemas/ManagedAgentsAnalyticsSeriesPoint' type: - array - 'null' series_group_by: description: >- Dimension series_groups is split by, echoing the request. Absent when no stacked series was asked for. type: string series_groups: description: >- One band per dimension value, for a stacked chart. Present only when series_group_by was requested. items: $ref: '#/components/schemas/ManagedAgentsAnalyticsSeriesGroup' type: array to: description: >- Exclusive end actually aggregated, widened outward to a bucket boundary. format: date-time type: string totals: $ref: '#/components/schemas/ManagedAgentsAnalyticsTotals' description: The whole window aggregated into one figure set. required: - family - family_version - scope - granularity - from - to - as_of - lag_seconds - totals - series - breakdowns type: object example: as_of: '2026-02-18T09:30:00.000Z' breakdowns: - dimension: example rows: - cache_read_tokens: 1 cache_write_tokens: 1 criterion_count: 1 criterion_fail_count: 1 criterion_not_applicable_count: 1 criterion_pass_count: 1 criterion_pass_rate: 1.5 duration_samples: 1 environment_failed: 1 environment_ready: 1 environment_starts: 1 environment_startup_ms: 1 evaluation_count: 1 evaluation_fail_count: 1 evaluation_not_applicable_count: 1 evaluation_pass_count: 1 evaluation_pass_rate: 1.5 event_count: 1 input_tokens: 1 label: example model_calls: 1 model_cost_micros: 1 output_tokens: 1 p50_ms: 1.5 p90_ms: 1.5 p95_ms: 1.5 session_completion_ms: 1 sessions: 1 tool_calls: 1 tool_completed: 1 tool_cost_micros: 1 tool_duration_ms: 1 tool_failed: 1 tool_in_flight: 1 total_cost_micros: 1 turns: 1 value: example family: example family_version: 1 from: '2026-02-18T09:30:00.000Z' granularity: hour lag_seconds: 1 scope: workspace series: - bucket_start: '2026-02-18T09:30:00.000Z' cache_read_tokens: 1 cache_write_tokens: 1 criterion_count: 1 criterion_fail_count: 1 criterion_not_applicable_count: 1 criterion_pass_count: 1 criterion_pass_rate: 1.5 duration_samples: 1 environment_failed: 1 environment_ready: 1 environment_starts: 1 environment_startup_ms: 1 evaluation_count: 1 evaluation_fail_count: 1 evaluation_not_applicable_count: 1 evaluation_pass_count: 1 evaluation_pass_rate: 1.5 event_count: 1 input_tokens: 1 model_calls: 1 model_cost_micros: 1 output_tokens: 1 p50_ms: 1.5 p90_ms: 1.5 p95_ms: 1.5 session_completion_ms: 1 sessions: 1 tool_calls: 1 tool_completed: 1 tool_cost_micros: 1 tool_duration_ms: 1 tool_failed: 1 tool_in_flight: 1 total_cost_micros: 1 turns: 1 series_group_by: example series_groups: - label: example points: - bucket_start: '2026-02-18T09:30:00.000Z' cache_read_tokens: 1 cache_write_tokens: 1 criterion_count: 1 criterion_fail_count: 1 criterion_not_applicable_count: 1 criterion_pass_count: 1 criterion_pass_rate: 1.5 duration_samples: 1 environment_failed: 1 environment_ready: 1 environment_starts: 1 environment_startup_ms: 1 evaluation_count: 1 evaluation_fail_count: 1 evaluation_not_applicable_count: 1 evaluation_pass_count: 1 evaluation_pass_rate: 1.5 event_count: 1 input_tokens: 1 model_calls: 1 model_cost_micros: 1 output_tokens: 1 p50_ms: 1.5 p90_ms: 1.5 p95_ms: 1.5 session_completion_ms: 1 sessions: 1 tool_calls: 1 tool_completed: 1 tool_cost_micros: 1 tool_duration_ms: 1 tool_failed: 1 tool_in_flight: 1 total_cost_micros: 1 turns: 1 value: example to: '2026-02-18T09:30:00.000Z' totals: cache_read_tokens: 1 cache_write_tokens: 1 criterion_count: 1 criterion_fail_count: 1 criterion_not_applicable_count: 1 criterion_pass_count: 1 criterion_pass_rate: 1.5 duration_samples: 1 environment_failed: 1 environment_ready: 1 environment_starts: 1 environment_startup_ms: 1 evaluation_count: 1 evaluation_fail_count: 1 evaluation_not_applicable_count: 1 evaluation_pass_count: 1 evaluation_pass_rate: 1.5 event_count: 1 input_tokens: 1 model_calls: 1 model_cost_micros: 1 output_tokens: 1 p50_ms: 1.5 p90_ms: 1.5 p95_ms: 1.5 session_completion_ms: 1 sessions: 1 tool_calls: 1 tool_completed: 1 tool_cost_micros: 1 tool_duration_ms: 1 tool_failed: 1 tool_in_flight: 1 total_cost_micros: 1 turns: 1 ManagedAgentsApiErrorInvalidRequest: type: object properties: code: type: string minLength: 1 description: Stable machine-readable error code. enum: - invalid_request message: type: string minLength: 1 description: Human-readable error message. details: $ref: '#/components/schemas/ManagedAgentsApiErrorDetails' description: >- Optional structured error details. Reserved transport fields are typed; code-specific fields remain forward compatible. required: - code - message additionalProperties: false description: Standard flat error response. ManagedAgentsApiErrorInvalidScope: type: object properties: code: type: string minLength: 1 description: Stable machine-readable error code. enum: - invalid_scope message: type: string minLength: 1 description: Human-readable error message. details: $ref: '#/components/schemas/ManagedAgentsApiErrorDetails' description: >- Optional structured error details. Reserved transport fields are typed; code-specific fields remain forward compatible. required: - code - message additionalProperties: false description: Standard flat error response. ManagedAgentsApiErrorUnauthorized: type: object properties: code: type: string minLength: 1 description: Stable machine-readable error code. enum: - unauthorized message: type: string minLength: 1 description: Human-readable error message. details: $ref: '#/components/schemas/ManagedAgentsApiErrorDetails' description: >- Optional structured error details. Reserved transport fields are typed; code-specific fields remain forward compatible. required: - code - message additionalProperties: false description: Standard flat error response. ManagedAgentsApiErrorForbidden: type: object properties: code: type: string minLength: 1 description: Stable machine-readable error code. enum: - forbidden message: type: string minLength: 1 description: Human-readable error message. details: $ref: '#/components/schemas/ManagedAgentsApiErrorDetails' description: >- Optional structured error details. Reserved transport fields are typed; code-specific fields remain forward compatible. required: - code - message additionalProperties: false description: Standard flat error response. ManagedAgentsApiErrorRateLimitExceeded: type: object properties: code: type: string minLength: 1 description: Stable machine-readable error code. enum: - rate_limit_exceeded message: type: string minLength: 1 description: Human-readable error message. details: $ref: '#/components/schemas/ManagedAgentsApiErrorDetails' description: >- Optional structured error details. Reserved transport fields are typed; code-specific fields remain forward compatible. required: - code - message additionalProperties: false description: Standard flat error response. ManagedAgentsApiErrorInternalError: type: object properties: code: type: string minLength: 1 description: Stable machine-readable error code. enum: - internal_error message: type: string minLength: 1 description: Human-readable error message. details: $ref: '#/components/schemas/ManagedAgentsApiErrorDetails' description: >- Optional structured error details. Reserved transport fields are typed; code-specific fields remain forward compatible. required: - code - message additionalProperties: false description: Standard flat error response. ManagedAgentsApiErrorInvariantViolation: type: object properties: code: type: string minLength: 1 description: Stable machine-readable error code. enum: - invariant_violation message: type: string minLength: 1 description: Human-readable error message. details: $ref: '#/components/schemas/ManagedAgentsApiErrorDetails' description: >- Optional structured error details. Reserved transport fields are typed; code-specific fields remain forward compatible. required: - code - message additionalProperties: false description: Standard flat error response. ManagedAgentsApiErrorBadGateway: type: object properties: code: type: string minLength: 1 description: Stable machine-readable error code. enum: - bad_gateway message: type: string minLength: 1 description: Human-readable error message. details: $ref: '#/components/schemas/ManagedAgentsApiErrorDetails' description: >- Optional structured error details. Reserved transport fields are typed; code-specific fields remain forward compatible. required: - code - message additionalProperties: false description: Standard flat error response. ManagedAgentsApiErrorAnalytics: type: object properties: code: type: string minLength: 1 description: Stable machine-readable error code. enum: - analytics message: type: string minLength: 1 description: Human-readable error message. details: $ref: '#/components/schemas/ManagedAgentsApiErrorDetails' description: >- Optional structured error details. Reserved transport fields are typed; code-specific fields remain forward compatible. required: - code - message additionalProperties: false description: Standard flat error response. ManagedAgentsApiErrorAnalyticsNotReady: type: object properties: code: type: string minLength: 1 description: Stable machine-readable error code. enum: - analytics_not_ready message: type: string minLength: 1 description: Human-readable error message. details: $ref: '#/components/schemas/ManagedAgentsApiErrorDetails' description: >- Optional structured error details. Reserved transport fields are typed; code-specific fields remain forward compatible. required: - code - message additionalProperties: false description: Standard flat error response. ManagedAgentsApiErrorManagedAgentsUnavailable: type: object properties: code: type: string minLength: 1 description: Stable machine-readable error code. enum: - managed_agents_unavailable message: type: string minLength: 1 description: Human-readable error message. details: $ref: '#/components/schemas/ManagedAgentsApiErrorDetails' description: >- Optional structured error details. Reserved transport fields are typed; code-specific fields remain forward compatible. required: - code - message additionalProperties: false description: Standard flat error response. ManagedAgentsApiErrorServiceUnavailable: type: object properties: code: type: string minLength: 1 description: Stable machine-readable error code. enum: - service_unavailable message: type: string minLength: 1 description: Human-readable error message. details: $ref: '#/components/schemas/ManagedAgentsApiErrorDetails' description: >- Optional structured error details. Reserved transport fields are typed; code-specific fields remain forward compatible. required: - code - message additionalProperties: false description: Standard flat error response. ManagedAgentsApiErrorGatewayTimeout: type: object properties: code: type: string minLength: 1 description: Stable machine-readable error code. enum: - gateway_timeout message: type: string minLength: 1 description: Human-readable error message. details: $ref: '#/components/schemas/ManagedAgentsApiErrorDetails' description: >- Optional structured error details. Reserved transport fields are typed; code-specific fields remain forward compatible. required: - code - message additionalProperties: false description: Standard flat error response. ManagedAgentsAnalyticsBreakdown: description: >- One dimension's groups for the requested window. Which dimensions appear is chosen by the family and the scope: a tenant or global read leads with workspace, since that is the question the narrower groupings cannot answer. properties: dimension: description: >- Dimension these rows are grouped by, e.g. tool_name, executing_agent_id, model, sandbox_provider, or compute_class. type: string rows: description: >- Groups ordered by the family's ranking measure, descending, capped by limit. items: $ref: '#/components/schemas/ManagedAgentsAnalyticsBreakdownRow' type: - array - 'null' required: - dimension - rows type: object example: dimension: example rows: - cache_read_tokens: 1 cache_write_tokens: 1 criterion_count: 1 criterion_fail_count: 1 criterion_not_applicable_count: 1 criterion_pass_count: 1 criterion_pass_rate: 1.5 duration_samples: 1 environment_failed: 1 environment_ready: 1 environment_starts: 1 environment_startup_ms: 1 evaluation_count: 1 evaluation_fail_count: 1 evaluation_not_applicable_count: 1 evaluation_pass_count: 1 evaluation_pass_rate: 1.5 event_count: 1 input_tokens: 1 label: example model_calls: 1 model_cost_micros: 1 output_tokens: 1 p50_ms: 1.5 p90_ms: 1.5 p95_ms: 1.5 session_completion_ms: 1 sessions: 1 tool_calls: 1 tool_completed: 1 tool_cost_micros: 1 tool_duration_ms: 1 tool_failed: 1 tool_in_flight: 1 total_cost_micros: 1 turns: 1 value: example ManagedAgentsAnalyticsSeriesPoint: description: >- Aggregated analytics measures for one window, group, or time bucket. Counters and sums are exact; p50_ms, p90_ms and p95_ms are approximate, derived from a histogram so they can be combined across time buckets and groups. properties: bucket_start: description: >- Inclusive start of this bucket, in UTC. Its width is the response's granularity. format: date-time type: string cache_read_tokens: description: Prompt tokens served from the provider's cache. format: int64 type: integer cache_write_tokens: description: Prompt tokens written into the provider's cache. format: int64 type: integer criterion_count: description: Immutable criterion verdict facts in the quality family. type: integer criterion_fail_count: description: Quality criterion verdicts that are fail. type: integer criterion_not_applicable_count: description: Quality criterion verdicts that are not applicable. type: integer criterion_pass_count: description: Quality criterion verdicts that are pass. type: integer criterion_pass_rate: description: >- Passed criteria divided by passed plus failed criteria. Null when that denominator is zero. type: - number - 'null' duration_samples: description: >- How many measurements the percentiles below were derived from. Compare with tool_calls or environment_starts to see the coverage. format: int64 type: integer environment_failed: description: >- Environment startups that failed or were cancelled before the sandbox ran. format: int64 type: integer environment_ready: description: Environment startups that reached a running sandbox. format: int64 type: integer environment_starts: description: >- Sandbox environment startups observed, including those still in progress. format: int64 type: integer environment_startup_ms: description: >- Summed time from the first provisioning transition to the first running transition, in milliseconds, over successful startups only. format: int64 type: integer evaluation_count: description: Immutable evaluation snapshots in the quality family. type: integer evaluation_fail_count: description: Quality evaluations whose parent verdict is fail. type: integer evaluation_not_applicable_count: description: Quality evaluations whose parent verdict is not applicable. type: integer evaluation_pass_count: description: Quality evaluations whose parent verdict is pass. type: integer evaluation_pass_rate: description: >- Passed evaluations divided by passed plus failed evaluations. Null when that denominator is zero. type: - number - 'null' event_count: description: >- Units of work recorded: every tool call, model call, and environment startup. format: int64 type: integer input_tokens: description: Prompt tokens billed in the window. format: int64 type: integer model_calls: description: >- Distinct model inference calls. A response persisted across several events is charged once. format: int64 type: integer model_cost_micros: description: Model spend in micro-USD (1,000,000 = 1 USD). format: int64 type: integer output_tokens: description: Completion tokens billed in the window. format: int64 type: integer p50_ms: description: >- Approximate median duration in milliseconds, interpolated within a fixed histogram bucket. Absent when nothing in the window reported a duration. type: number p90_ms: description: >- Approximate 90th percentile duration in milliseconds, interpolated within a fixed histogram bucket. Absent when nothing in the window reported a duration. type: number p95_ms: description: >- Approximate 95th percentile duration in milliseconds, interpolated within a fixed histogram bucket. Absent when nothing in the window reported a duration. type: number session_completion_ms: description: >- Sum of creation-to-first-recorded-successful-completion durations in milliseconds. Divide by sessions for the mean in the completion family. Earlier sessions without a recorded completion milestone are excluded. type: integer sessions: description: >- For managed_agent_completion: root sessions with a first recorded successful completion in the window, bucketed by completion time. For other families: sessions counted once in the bucket they started in. format: int64 type: integer tool_calls: description: >- Tool invocations in the window. A call is counted once, in the bucket it started in, whether or not it has finished. format: int64 type: integer tool_completed: description: Tool calls whose result reported success. format: int64 type: integer tool_cost_micros: description: >- Directly attributable tool spend in micro-USD, such as a metered built-in tool. format: int64 type: integer tool_duration_ms: description: >- Summed measured duration of every tool call that reported one, in milliseconds. Model thinking time is not included. format: int64 type: integer tool_failed: description: >- Tool calls whose result reported an error, plus invocations that failed before any result was recorded. format: int64 type: integer tool_in_flight: description: >- Tool calls with no result yet. A call that never receives one stays here permanently. format: int64 type: integer total_cost_micros: description: model_cost_micros plus tool_cost_micros. format: int64 type: integer turns: description: >- Model turns, counted once each from the runtime's own turn grouping. Only the managed_agent_turn family records these; divided by sessions it is turns per session. format: int64 type: integer required: - bucket_start - sessions - turns - event_count - tool_calls - tool_completed - tool_failed - tool_in_flight - tool_duration_ms - model_calls - input_tokens - output_tokens - cache_read_tokens - cache_write_tokens - model_cost_micros - tool_cost_micros - total_cost_micros - environment_starts - environment_ready - environment_failed - environment_startup_ms - duration_samples type: object example: bucket_start: '2026-02-18T09:30:00.000Z' cache_read_tokens: 1 cache_write_tokens: 1 criterion_count: 1 criterion_fail_count: 1 criterion_not_applicable_count: 1 criterion_pass_count: 1 criterion_pass_rate: 1.5 duration_samples: 1 environment_failed: 1 environment_ready: 1 environment_starts: 1 environment_startup_ms: 1 evaluation_count: 1 evaluation_fail_count: 1 evaluation_not_applicable_count: 1 evaluation_pass_count: 1 evaluation_pass_rate: 1.5 event_count: 1 input_tokens: 1 model_calls: 1 model_cost_micros: 1 output_tokens: 1 p50_ms: 1.5 p90_ms: 1.5 p95_ms: 1.5 session_completion_ms: 1 sessions: 1 tool_calls: 1 tool_completed: 1 tool_cost_micros: 1 tool_duration_ms: 1 tool_failed: 1 tool_in_flight: 1 total_cost_micros: 1 turns: 1 ManagedAgentsAnalyticsSeriesGroup: description: >- One band of a stacked series. Percentiles are omitted on these points, because a stacked chart plots additive counts. properties: label: description: >- Human-readable name for value, resolved when the request is served, on the same terms as a breakdown row's label. type: string points: description: >- This line's buckets, ascending. Buckets where this group recorded nothing are omitted, so a stacked chart must zero-fill them. items: $ref: '#/components/schemas/ManagedAgentsAnalyticsSeriesPoint' type: - array - 'null' value: description: >- The dimension's value for this line, and the value to pass back as a filter. Empty means the dimension was not recorded on the underlying work. type: string required: - value - points type: object example: label: example points: - bucket_start: '2026-02-18T09:30:00.000Z' cache_read_tokens: 1 cache_write_tokens: 1 criterion_count: 1 criterion_fail_count: 1 criterion_not_applicable_count: 1 criterion_pass_count: 1 criterion_pass_rate: 1.5 duration_samples: 1 environment_failed: 1 environment_ready: 1 environment_starts: 1 environment_startup_ms: 1 evaluation_count: 1 evaluation_fail_count: 1 evaluation_not_applicable_count: 1 evaluation_pass_count: 1 evaluation_pass_rate: 1.5 event_count: 1 input_tokens: 1 model_calls: 1 model_cost_micros: 1 output_tokens: 1 p50_ms: 1.5 p90_ms: 1.5 p95_ms: 1.5 session_completion_ms: 1 sessions: 1 tool_calls: 1 tool_completed: 1 tool_cost_micros: 1 tool_duration_ms: 1 tool_failed: 1 tool_in_flight: 1 total_cost_micros: 1 turns: 1 value: example ManagedAgentsAnalyticsTotals: description: >- Aggregated analytics measures for one window, group, or time bucket. Counters and sums are exact; p50_ms, p90_ms and p95_ms are approximate, derived from a histogram so they can be combined across time buckets and groups. properties: cache_read_tokens: description: Prompt tokens served from the provider's cache. format: int64 type: integer cache_write_tokens: description: Prompt tokens written into the provider's cache. format: int64 type: integer criterion_count: description: Immutable criterion verdict facts in the quality family. type: integer criterion_fail_count: description: Quality criterion verdicts that are fail. type: integer criterion_not_applicable_count: description: Quality criterion verdicts that are not applicable. type: integer criterion_pass_count: description: Quality criterion verdicts that are pass. type: integer criterion_pass_rate: description: >- Passed criteria divided by passed plus failed criteria. Null when that denominator is zero. type: - number - 'null' duration_samples: description: >- How many measurements the percentiles below were derived from. Compare with tool_calls or environment_starts to see the coverage. format: int64 type: integer environment_failed: description: >- Environment startups that failed or were cancelled before the sandbox ran. format: int64 type: integer environment_ready: description: Environment startups that reached a running sandbox. format: int64 type: integer environment_starts: description: >- Sandbox environment startups observed, including those still in progress. format: int64 type: integer environment_startup_ms: description: >- Summed time from the first provisioning transition to the first running transition, in milliseconds, over successful startups only. format: int64 type: integer evaluation_count: description: Immutable evaluation snapshots in the quality family. type: integer evaluation_fail_count: description: Quality evaluations whose parent verdict is fail. type: integer evaluation_not_applicable_count: description: Quality evaluations whose parent verdict is not applicable. type: integer evaluation_pass_count: description: Quality evaluations whose parent verdict is pass. type: integer evaluation_pass_rate: description: >- Passed evaluations divided by passed plus failed evaluations. Null when that denominator is zero. type: - number - 'null' event_count: description: >- Units of work recorded: every tool call, model call, and environment startup. format: int64 type: integer input_tokens: description: Prompt tokens billed in the window. format: int64 type: integer model_calls: description: >- Distinct model inference calls. A response persisted across several events is charged once. format: int64 type: integer model_cost_micros: description: Model spend in micro-USD (1,000,000 = 1 USD). format: int64 type: integer output_tokens: description: Completion tokens billed in the window. format: int64 type: integer p50_ms: description: >- Approximate median duration in milliseconds, interpolated within a fixed histogram bucket. Absent when nothing in the window reported a duration. type: number p90_ms: description: >- Approximate 90th percentile duration in milliseconds, interpolated within a fixed histogram bucket. Absent when nothing in the window reported a duration. type: number p95_ms: description: >- Approximate 95th percentile duration in milliseconds, interpolated within a fixed histogram bucket. Absent when nothing in the window reported a duration. type: number session_completion_ms: description: >- Sum of creation-to-first-recorded-successful-completion durations in milliseconds. Divide by sessions for the mean in the completion family. Earlier sessions without a recorded completion milestone are excluded. type: integer sessions: description: >- For managed_agent_completion: root sessions with a first recorded successful completion in the window, bucketed by completion time. For other families: sessions counted once in the bucket they started in. format: int64 type: integer tool_calls: description: >- Tool invocations in the window. A call is counted once, in the bucket it started in, whether or not it has finished. format: int64 type: integer tool_completed: description: Tool calls whose result reported success. format: int64 type: integer tool_cost_micros: description: >- Directly attributable tool spend in micro-USD, such as a metered built-in tool. format: int64 type: integer tool_duration_ms: description: >- Summed measured duration of every tool call that reported one, in milliseconds. Model thinking time is not included. format: int64 type: integer tool_failed: description: >- Tool calls whose result reported an error, plus invocations that failed before any result was recorded. format: int64 type: integer tool_in_flight: description: >- Tool calls with no result yet. A call that never receives one stays here permanently. format: int64 type: integer total_cost_micros: description: model_cost_micros plus tool_cost_micros. format: int64 type: integer turns: description: >- Model turns, counted once each from the runtime's own turn grouping. Only the managed_agent_turn family records these; divided by sessions it is turns per session. format: int64 type: integer required: - sessions - turns - event_count - tool_calls - tool_completed - tool_failed - tool_in_flight - tool_duration_ms - model_calls - input_tokens - output_tokens - cache_read_tokens - cache_write_tokens - model_cost_micros - tool_cost_micros - total_cost_micros - environment_starts - environment_ready - environment_failed - environment_startup_ms - duration_samples type: object example: cache_read_tokens: 1 cache_write_tokens: 1 criterion_count: 1 criterion_fail_count: 1 criterion_not_applicable_count: 1 criterion_pass_count: 1 criterion_pass_rate: 1.5 duration_samples: 1 environment_failed: 1 environment_ready: 1 environment_starts: 1 environment_startup_ms: 1 evaluation_count: 1 evaluation_fail_count: 1 evaluation_not_applicable_count: 1 evaluation_pass_count: 1 evaluation_pass_rate: 1.5 event_count: 1 input_tokens: 1 model_calls: 1 model_cost_micros: 1 output_tokens: 1 p50_ms: 1.5 p90_ms: 1.5 p95_ms: 1.5 session_completion_ms: 1 sessions: 1 tool_calls: 1 tool_completed: 1 tool_cost_micros: 1 tool_duration_ms: 1 tool_failed: 1 tool_in_flight: 1 total_cost_micros: 1 turns: 1 ManagedAgentsApiErrorDetails: type: object properties: field: description: >- Request field or header responsible for the error, when one can be identified. type: string issues: description: >- Boundary-validation failures as path-prefixed human-readable messages. type: array items: type: string requestId: description: Request correlation identifier for support and log lookup. type: string retryable: description: >- Server advice about failure transience. `true` means transient, `false` means non-transient, and absence gives no advice. Automatic replay is allowed only when this field is not `false` and the operation-specific retry and idempotency contract permits replay. type: boolean additionalProperties: {} description: >- Optional structured error details. Reserved transport fields are typed; code-specific fields remain forward compatible. ManagedAgentsAnalyticsBreakdownRow: description: >- Aggregated analytics measures for one window, group, or time bucket. Counters and sums are exact; p50_ms, p90_ms and p95_ms are approximate, derived from a histogram so they can be combined across time buckets and groups. properties: cache_read_tokens: description: Prompt tokens served from the provider's cache. format: int64 type: integer cache_write_tokens: description: Prompt tokens written into the provider's cache. format: int64 type: integer criterion_count: description: Immutable criterion verdict facts in the quality family. type: integer criterion_fail_count: description: Quality criterion verdicts that are fail. type: integer criterion_not_applicable_count: description: Quality criterion verdicts that are not applicable. type: integer criterion_pass_count: description: Quality criterion verdicts that are pass. type: integer criterion_pass_rate: description: >- Passed criteria divided by passed plus failed criteria. Null when that denominator is zero. type: - number - 'null' duration_samples: description: >- How many measurements the percentiles below were derived from. Compare with tool_calls or environment_starts to see the coverage. format: int64 type: integer environment_failed: description: >- Environment startups that failed or were cancelled before the sandbox ran. format: int64 type: integer environment_ready: description: Environment startups that reached a running sandbox. format: int64 type: integer environment_starts: description: >- Sandbox environment startups observed, including those still in progress. format: int64 type: integer environment_startup_ms: description: >- Summed time from the first provisioning transition to the first running transition, in milliseconds, over successful startups only. format: int64 type: integer evaluation_count: description: Immutable evaluation snapshots in the quality family. type: integer evaluation_fail_count: description: Quality evaluations whose parent verdict is fail. type: integer evaluation_not_applicable_count: description: Quality evaluations whose parent verdict is not applicable. type: integer evaluation_pass_count: description: Quality evaluations whose parent verdict is pass. type: integer evaluation_pass_rate: description: >- Passed evaluations divided by passed plus failed evaluations. Null when that denominator is zero. type: - number - 'null' event_count: description: >- Units of work recorded: every tool call, model call, and environment startup. format: int64 type: integer input_tokens: description: Prompt tokens billed in the window. format: int64 type: integer label: description: >- Human-readable name for value, resolved when the request is served. Present for identifier dimensions (agent, agent version, environment); absent for dimensions that are already readable, and for an identifier whose subject has since been deleted. type: string model_calls: description: >- Distinct model inference calls. A response persisted across several events is charged once. format: int64 type: integer model_cost_micros: description: Model spend in micro-USD (1,000,000 = 1 USD). format: int64 type: integer output_tokens: description: Completion tokens billed in the window. format: int64 type: integer p50_ms: description: >- Approximate median duration in milliseconds, interpolated within a fixed histogram bucket. Absent when nothing in the window reported a duration. type: number p90_ms: description: >- Approximate 90th percentile duration in milliseconds, interpolated within a fixed histogram bucket. Absent when nothing in the window reported a duration. type: number p95_ms: description: >- Approximate 95th percentile duration in milliseconds, interpolated within a fixed histogram bucket. Absent when nothing in the window reported a duration. type: number session_completion_ms: description: >- Sum of creation-to-first-recorded-successful-completion durations in milliseconds. Divide by sessions for the mean in the completion family. Earlier sessions without a recorded completion milestone are excluded. type: integer sessions: description: >- For managed_agent_completion: root sessions with a first recorded successful completion in the window, bucketed by completion time. For other families: sessions counted once in the bucket they started in. format: int64 type: integer tool_calls: description: >- Tool invocations in the window. A call is counted once, in the bucket it started in, whether or not it has finished. format: int64 type: integer tool_completed: description: Tool calls whose result reported success. format: int64 type: integer tool_cost_micros: description: >- Directly attributable tool spend in micro-USD, such as a metered built-in tool. format: int64 type: integer tool_duration_ms: description: >- Summed measured duration of every tool call that reported one, in milliseconds. Model thinking time is not included. format: int64 type: integer tool_failed: description: >- Tool calls whose result reported an error, plus invocations that failed before any result was recorded. format: int64 type: integer tool_in_flight: description: >- Tool calls with no result yet. A call that never receives one stays here permanently. format: int64 type: integer total_cost_micros: description: model_cost_micros plus tool_cost_micros. format: int64 type: integer turns: description: >- Model turns, counted once each from the runtime's own turn grouping. Only the managed_agent_turn family records these; divided by sessions it is turns per session. format: int64 type: integer value: description: >- The dimension's value for this group, and the value to pass back as a filter. Empty means the dimension was not recorded on the underlying work. type: string required: - value - sessions - turns - event_count - tool_calls - tool_completed - tool_failed - tool_in_flight - tool_duration_ms - model_calls - input_tokens - output_tokens - cache_read_tokens - cache_write_tokens - model_cost_micros - tool_cost_micros - total_cost_micros - environment_starts - environment_ready - environment_failed - environment_startup_ms - duration_samples type: object example: cache_read_tokens: 1 cache_write_tokens: 1 criterion_count: 1 criterion_fail_count: 1 criterion_not_applicable_count: 1 criterion_pass_count: 1 criterion_pass_rate: 1.5 duration_samples: 1 environment_failed: 1 environment_ready: 1 environment_starts: 1 environment_startup_ms: 1 evaluation_count: 1 evaluation_fail_count: 1 evaluation_not_applicable_count: 1 evaluation_pass_count: 1 evaluation_pass_rate: 1.5 event_count: 1 input_tokens: 1 label: example model_calls: 1 model_cost_micros: 1 output_tokens: 1 p50_ms: 1.5 p90_ms: 1.5 p95_ms: 1.5 session_completion_ms: 1 sessions: 1 tool_calls: 1 tool_completed: 1 tool_cost_micros: 1 tool_duration_ms: 1 tool_failed: 1 tool_in_flight: 1 total_cost_micros: 1 turns: 1 value: example securitySchemes: bearerAuth: type: http scheme: bearer description: A Recursion API key, created in the console under API keys. ````