@launchdarkly/observability + @launchdarkly/session-replay
    Preparing search index...

    Type Alias ObserveOptions

    ObserveOptions: CommonOptions & {
        captureQueryParams?: string[] | false;
        captureReferrer?: boolean;
        consoleMethodsToRecord?: ConsoleMethods[];
        disableConsoleRecording?: boolean;
        enableFlagEvaluationSpans?: boolean;
        enableLongtaskRecording?: boolean;
        enablePerformanceRecording?: boolean;
        enablePromisePatch?: boolean;
        enableReportingObserver?: boolean;
        environment?: "development" | "staging" | "production" | string;
        flagExposureDedupeMaxSize?: number;
        flagExposureDedupeWindowMillis?: number;
        networkRecording?: boolean | NetworkRecordingOptions;
        otel?: OtelOptions & { otlpEndpoint?: string };
        productAnalytics?: boolean | ProductAnalyticsEvents;
        recordLaunchDarklySdkLogs?: boolean;
        reportConsoleErrors?: boolean;
        tracingOrigins?: boolean | (string | RegExp)[];
    }

    Type declaration

    • OptionalcaptureQueryParams?: string[] | false

      Query parameters to read from the page URL and add as attributes to all telemetry, as if passed to LDObserve.setAttributes.

      When the SDK initializes, each listed parameter present in location.search becomes an attribute named after the parameter (e.g. utm_campaign) on every span (including document load, page view and fetch/XHR spans), log, error and metric for the rest of the page load, including after SPA navigation drops it from the URL.

      Last touch: when an SPA navigation lands on a URL with any listed parameter, the captured values are replaced by that URL's values, and previously captured keys missing from it are removed, so the set always reflects one landing. A navigation without listed parameters keeps the current values. Spans already started keep the values they had.

      Values are kept in memory only; nothing is written to storage or cookies, so a later full page load captures again from its own URL.

      Only the listed keys are captured, so other query parameters (which may hold personal data such as emails or tokens) are never recorded this way. The default list contains UTM parameters only. A name ending in * matches every parameter with that prefix (e.g. 'utm_*').

      Keys set with setAttributes win: capture never replaces or removes a key the app set itself. removeAttributes removes a captured value (a later landing with that parameter captures it again). Keys in the namespaces the SDK sets itself (highlight.*, launchdarkly.*, session.*, context.contextKeys.*) are ignored.

      A custom list replaces the default. Set to false or [] to turn capture off.

      ['utm_source', 'utm_medium', 'utm_campaign', 'utm_term', 'utm_content']
      
      captureQueryParams: ['utm_source', 'utm_campaign', 'gclid']
      
    • OptionalcaptureReferrer?: boolean

      Specifies whether to record the page that linked to this one. When the SDK initializes and document.referrer is a different origin, it is added as the landing.referrer attribute to all telemetry for the rest of the page load, as if passed to LDObserve.setAttributes.

      Only the origin and path are kept: the query string, fragment and any credentials are stripped so tokens or personal data in the referrer URL are not recorded. An empty or same-origin referrer sets nothing, and the value is not updated on SPA navigation.

      true
      
    • OptionalconsoleMethodsToRecord?: ConsoleMethods[]

      Specifies which console methods to record. The value here will be ignored if disabledConsoleRecording is true.

      All console methods.
      
      consoleMethodsToRecord: ['log', 'info', 'error']
      
    • OptionaldisableConsoleRecording?: boolean

      Specifies whether the SDK will record console messages.

      false
      
    • OptionalenableFlagEvaluationSpans?: boolean

      Specifies whether to emit a feature_flag exposure span for each feature flag evaluation made by the LaunchDarkly client.

      Each evaluation span is ingested as a trace, so evaluation-heavy pages can produce a large trace volume. Set to false to stop emitting evaluation spans while keeping the rest of the plugin's telemetry (for example web vitals, errors and logs). Other plugins' evaluation hooks still run.

      When enabled, flagExposureDedupeWindowMillis can be used instead to reduce the volume without turning the spans off entirely.

      true
      
    • OptionalenableLongtaskRecording?: boolean

      Specifies whether to record main-thread longtask entries (>50ms) as long_task.duration histogram samples.

      true
      
    • OptionalenablePerformanceRecording?: boolean

      Specifies whether to record performance metrics (e.g. FPS, device memory).

      true
      
    • OptionalenablePromisePatch?: boolean

      Specifies whether window.Promise should be patched to record the stack trace of promise rejections.

      true
      
    • OptionalenableReportingObserver?: boolean

      Specifies whether to subscribe to the browser Reporting API and emit CSP/intervention reports as errors and deprecation reports as warn logs.

      true
      
    • Optionalenvironment?: "development" | "staging" | "production" | string

      Specifies the environment your application is running in. This is useful to distinguish whether your session was recorded on localhost or in production.

      'production'
      
    • OptionalflagExposureDedupeMaxSize?: number

      The maximum number of unique feature flag exposure keys tracked for deduplication at once. When exceeded, the least recently recorded keys are evicted to bound memory usage.

      2000
      
    • OptionalflagExposureDedupeWindowMillis?: number

      The time window, in milliseconds, during which repeated feature flag evaluations that resolve to the same result are deduplicated, so that only a single feature_flag exposure span is emitted per unique (flag key, value, variation, reason, context) within the window.

      This is useful for reducing exposure volume caused by frequent re-evaluations (for example, React re-renders).

      Set to 0 (the default) to disable deduplication and emit an exposure for every evaluation. Set a positive value to enable it.

      0 (disabled)
      
    • OptionalnetworkRecording?: boolean | NetworkRecordingOptions

      Specifies how and what the SDK records from network requests and responses.

    • Optionalotel?: OtelOptions & { otlpEndpoint?: string }

      OTLP options for OpenTelemetry tracing. Instrumentations are enabled by default.

    • OptionalproductAnalytics?: boolean | ProductAnalyticsEvents

      Specifies whether to record product analytics events.

      true
      
    • OptionalrecordLaunchDarklySdkLogs?: boolean

      Specifies whether console output written by the LaunchDarkly SDK's own logger (lines prefixed [LaunchDarkly], e.g. stream connection open/close messages) is recorded as logs. Those lines describe the SDK rather than your application, and the streaming URL they contain embeds the evaluation context, so they are skipped by default.

      false
      
    • OptionalreportConsoleErrors?: boolean

      Specifies whether the SDK will report console.error invocations as Errors.

      false
      
    • OptionaltracingOrigins?: boolean | (string | RegExp)[]

      Specifies where the backend of the app lives. If specified, the SDK will attach the traceparent header to outgoing requests whose destination URLs match a substring or regexp from this list, so that backend errors can be linked back to the session. If 'true' is specified, all requests to the current domain will be matched.

      tracingOrigins: ['localhost', /^//, 'backend.myapp.com']