Logs, traces, metrics, and error tracking for production systems.
Datadog patched a tracing vulnerability across every Python tracer line it maintains, while Sentry expanded agent tracing to more AI frameworks and Datadog shipped a new AI SDLC view of coding-agent spend.
Datadog APM fixed a tracing security vulnerability and closed an AI Guard gap that let LangChain model responses bypass unsafe-output evaluation, while Sentry added filtering and sorting of agent conversations by totals like cost, tokens, and tool calls.
Explore > Agents now supports filtering and sorting conversations on whole-conversation aggregates like conversation.totalCost, conversation.totalTokens, conversation.toolCalls, and conversation.errors, rather than only individual spans. Column headers are sortable, so conversations can be ordered by cost or errors to surface ones needing attention.
Datadog's Python tracer fixed a tracing security flaw and stopped leaking AI Guard credentials into logs and spans, while Sentry's JavaScript SDK folded its bundler plugins into the SDK repo with a Node 20.19 floor and Datadog added LLM Observability sampling rules and Azure Durable Functions tracing.
Labels like env=production or team=platform can now be created with a color and optional description and assigned to datasets, dashboards, monitors, and notifiers, then used to filter resource lists with ANY or ALL matching, with selections preserved in the page URL. The MCP server exposes label names and descriptions alongside resource metadata to give agents service, environment, and ownership context.
Datadog's Agent now denies Private Action Runner access to core Kubernetes APIs by default, while the JavaScript tracer added WebdriverIO failure videos and the Browser SDK began collecting WebSocket vitals.
Datadog introduced Scout 1.0, its first AI model specialized for investigating production issues, which powers its AI SRE Bits Investigation. Trained on GLM-5.3 Flash with reinforcement learning over historical incidents and telemetry simulations, Scout 1.0 medium matches GPT-6.1 Sol high and Claude Opus 5.5 high at 3-8x lower estimated inference cost, with the xhigh mode matching Fable xhigh at 4x lower cost.
Datadog examines how the legacy device plugin model limits GPU utilization through fixed MIG partition profiles, opaque GPU counts that can't express memory or topology requirements, and a lack of scheduler-managed capacity accounting. The post covers the three mechanisms Kubernetes Dynamic Resource Allocation uses to raise utilization and how Datadog GPU Monitoring can track the resulting cost savings.
Operations Monitoring, included with RUM at no additional cost, lets teams define application-controlled steps like "process payment" with their own starts, outcomes, and durations, then track availability and latency independently using out-of-the-box metrics rum.measure.operation and rum.measure.operation.duration. Teams can set separate availability and latency SLOs per operation and use Bits AI to investigate degradations.
The TurboModule Context integration records calls to React Native New Architecture native modules, capturing the module and method active during native crashes, call counts, error counts, and durations per module and method on transactions and root spans, and breadcrumbs for slow asynchronous calls. It is enabled by default but wraps only the Sentry module; register your own modules via the modules option, and set enableLegacyNativeModules: true on the Old Architecture to record all registered NativeModules automatically.
Sentry's Android SDK gained experimental Jetpack Navigation 3 instrumentation, letting SentryNavEffect capture navigation transactions, breadcrumbs, screen names, and back stack context from NavDisplay.
Starting with Sentry Android SDK 8.60.0, SentryNavEffect used alongside NavDisplay records navigation transactions, breadcrumbs, screen names, and back stack context via SentryNavOptions. The initial release is experimental and covers single-stack apps, adding back stack context to what the Nav2 integration already produced.
Sentry's JavaScript SDK added opt-in Cloudflare Cron Trigger monitoring and automatic Neon database spans, while the Python SDK deprecated send_default_pii and transaction profiling and improved inbound filtering and body scrubbing.
Datadog describes building a service criticality system that combines incident postmortem history, APM/RUM error traces, traffic volume, and on-call monitors into high, low, or unknown levels with a confidence score. The postmortem signal used an LLM to extract involved services from several hundred SEV-1 and SEV-2 postmortems, validated against Software Catalog.
The Sentry Rust SDK now captures native crashes like segmentation faults as minidumps across Linux, macOS, and Windows, covering failures in unsafe code and FFI calls that never trigger a panic.
The SDK now captures native crashes such as segmentation faults as minidumps and sends them to Sentry on Linux, macOS, and Windows, covering crashes in unsafe code and FFI calls into C libraries that occur without a panic. Enabled by compiling the sentry crate with the minidump feature and adding MinidumpIntegration at init.
Sentry Logs added regex pattern matching in Explore, Dash0 Operator deprecated the container.image.tag resource attribute ahead of its removal, and Datadog's Python tracer patched a tracing security vulnerability.
Sentry Logs search now supports regular expressions, letting you filter any string attribute by pattern rather than exact text via a "matches regex" operator in Explore → Logs. Patterns use RE2 syntax, are case-sensitive by default, and are capped at 64 characters.
APM
Sentry
Dash0
Axiom