← Back to all sparks
B

Bugsnag

INFRA · APIS
Velocity2.5

Bugsnag's Correlated Events links errors across users, devices, sessions, and traces — debugging context goes cross-service.

error-trackingcorrelated-eventsmcpmobile-monitoringflutterobservability
Current state
Bugsnag is shipping monthly release notes with steady improvements across its error monitoring and performance tooling. The standout move in this cycle is Correlated Events, which surfaces related events sharing the same user, device, session, or trace from the event view — enabling cross-service pattern detection rather than single-service point debugging. The MCP server (with OAuth 2.0 support) and monthly Flutter SDK improvements fill out a product that is broadening its observability surface across mobile, backend, and AI agent workflows.
Where it's heading
Bugsnag is moving from single-service error tracking toward cross-service observability: Correlated Events, trace correlation, and the MCP server all point at workflows where an error in one service is causally linked to events in others. The Flutter SDK parity work (CPU, memory, rendering metrics) reflects a deliberate effort to close the feature gap between mobile and web monitoring. SDK Source-Side Discard Enforcement (remote rule push to SDK) reduces bandwidth and processing overhead at scale — a signal that larger enterprise deployments are a growth priority.
Prediction
Correlated Events will likely expand to support more correlation dimensions — custom attributes, error clusters — and the MCP server toolset will grow to include mutation tools (creating alerts, snoozing errors) rather than just read-only access. The Performance Score metric will gain more visibility and dashboard integrations.

Recent moves

  1. 14d ago

    Bugsnag August: Remote SDK Source-Side Discard Enforcement goes live

    Source-Side Discard Enforcement pushing dashboard rules directly to upgraded Android and iOS SDKs reduces network and processing overhead at scale — a real improvement for high-volume apps where server-side filtering was a throughput bottleneck.

    View source ↗
  2. 1mo ago

    Bugsnag July: Custom session windows and per-session CPU/memory metrics

    Custom session windows via the SDK API let teams define meaningful performance measurement boundaries (e.g., user flows rather than page loads), and per-session CPU/memory aggregates add performance-monitoring depth that previously required separate tooling.

    View source ↗
  3. 2mo ago

    Bugsnag June: Side-by-side span comparison and Vue Router v5 support

    Side-by-side span instance comparison is a targeted debugging improvement for intermittent performance regressions where the difference between a fast and a slow execution is the diagnostic signal — incremental but genuinely useful for performance investigation.

    View source ↗
  4. 3mo ago

    Bugsnag May: iOS startup pre-main view, MCP Server OAuth 2.0, Flutter system metrics

    Pre-main aggregate view for iOS App Starts surfaces startup performance at the phase that's hardest to instrument manually. MCP Server OAuth 2.0 removes the manual API-token requirement that was the friction point for AI agent integration — a usability improvement that matters for teams adopting MCP-based workflows.

    View source ↗
  5. 4mo ago

    Bugsnag April: Correlated Events links errors across users, devices, sessions, and traces

    ⚡ SPARK

    Correlated Events is the most directional move in this batch: surfacing related BugSnag events across user, device, session, and trace dimensions from a single event view turns Bugsnag from a single-error debugger into a cross-service pattern detector — a meaningful capability jump.

    View source ↗
  6. 4mo ago

    Bugsnag March: Android App Hang detection, HTTP error reporting in Android and iOS SDKs

    App Hang detection on Android fills the gap between ANR events and shorter freezes, and HTTP error reporting in Android and iOS SDKs brings network-layer errors into Bugsnag's event stream — both expanding the error surface that Bugsnag can monitor without custom instrumentation.

    View source ↗