Updated the default query string obfuscation regex. Also removed axios from test dependencies.
Datadog dd-trace-js
npx @buildinternet/releases get datadog-dd-trace-jsThe default regex used to obfuscate query strings is updated. Additionally, axios was removed from the test suite.
Adds OpenTelemetry consistent probability sampling state to spans exported through OTLP and OTel sampling support in distributed tracing, plus remote evaluators and BaseEvaluator support for LLM Observability Experiments. Dynamic Instrumentation now redacts sensitive identifiers in log templates and capture expressions.
LLM Observability gains remote evaluators and BaseEvaluator support for Experiments, and OpenTelemetry spans exported through OTLP now carry consistent probability sampling state with OTel sampling support in distributed tracing. Dynamic Instrumentation adds an evaluation time budget and redacts sensitive identifiers in log templates and capture expressions.
Test Optimization failure media uploads are now opt-out, and AppSec/AI Guard supports anthropic stream responses. Plus fixes across LLM Observability prompt rendering, oracledb Easy Connect URL parsing, undici duplicate fetch tracing, and profiling shutdown exports.
Test Optimization failure media uploads are now opt-out, and LLM Observability emits gen_ai.* APM tags when LLM Observability is disabled. Fixed duplicate tracing of wrapped undici fetch requests, profiler config updates during in-flight shutdown exports, and OTLP adoption markers added to trace payloads.
LLM Observability now exposes versioned prompt config and supports message placeholders in managed prompts. Fixed a ws crash and mistagging on sends without an attached socket, alongside a batch of Test Optimization fixes for last-failed history, ATR budgets, and Playwright discovery.
LLM Observability now exposes versioned prompt config and supports message placeholders in managed prompts, alongside new Turbopack support for hookless rewriter targets. Fixed a ws crash when sending without an attached socket plus a dozen Test Optimization fixes covering skipped sessions, preserved run errors, and last-failed history.
Allocation profiling and the auto OOM heap limit are now enabled by default, and remote-config can activate or deactivate profiling. Added Postgres.js and Supabase tracing, Postgres DBM propagation, LLM Observability prompt management and gen_ai.* span attributes, and dynamic ATR support across Jest, Mocha, Vitest, Cypress, Playwright, Cucumber, and WebdriverIO.
Allocation profiling and auto OOM heap limit are now enabled by default, and the release adds a Postgres.js tracing integration, Supabase tracing, prompt management for LLM Observability, and dynamic ATR support across Jest, Mocha, Cypress, Playwright, Vitest, Cucumber, and WebdriverIO. Also fixed a thick mode crash for oracledb connection strings, a Dynamic Instrumentation crash on oversized payloads, and restored v5 framework compatibility in Test Optimization.
New instrumentation covers Next.js server bundles built with Turbopack, Azure Durable Functions orchestration triggers, and ai-vercel streaming for AI Guard, plus meter provider shutdown for OpenTelemetry. Fixes include span finalization through the azure-cosmos plugin lifecycle and native ESM export resolution in esbuild.
Adds instrumentation for Next.js server bundles built with Turbopack, Azure Functions Durable orchestration triggers, and AI Guard streaming for ai-vercel, plus a meter provider shutdown for OpenTelemetry. Fixes include finalizing azure-cosmos spans through the plugin lifecycle, restoring dd-trace-api tracer compatibility, and deduping rewriter hooks by version range and file.
Adds trace discarding through DD_TRACE_SAMPLING_RULES in scope, plus proxy environment variable support and port configuration validation. Fixed SQL and command evidence parsing in AppSec, guarded invalid text map carriers, used proxy for OTLP export, and preserved Jest logger module realms and delayed starts in Test Optimization.
APM now discards traces through DD_TRACE_SAMPLING_RULES, and proxy environment variables are respected for tracing and OTLP export. Also validates port configuration and guards invalid text map carriers at input boundaries.
Data pipeline-backed agentless mode now supports APM, Crash Tracking, and Dynamic Instrumentation, and graphql-jit executions are traced with reduced inactive resolver overhead. Fixes include LLM Observability unbounded memory growth in the vercel ai sdk plugin and numerous Test Optimization stability and compatibility improvements.
Introduces a data pipeline-backed agentless mode for APM, Crash Tracking, and Dynamic Instrumentation, enabling direct intake without a local agent. Fixes include resolving unbounded memory growth in the Vercel AI SDK LLM Observability plugin and support for Mocha 12 and Vitest 5 in Test Optimization, plus broader correctness fixes across graphql, LLM Observability, and Serverless.
AI Guard now redacts sensitive data, and Test Optimization submits Bunyan and Pino logs agentlessly. Also added DD_TRACE_HTTP_CLIENT_ERROR_STATUSES, LLM Observability experiment concurrency, records in datasets, project overrides, opt-out for all integrations, and a recordException API; fixed stale blocking actions, mariadb 3.5.3 support, and profiling label/web tag issues along with Test Optimization flush/worker fixes.
Sensitive data redaction via AI Guard and a new recordException API for tracer errors. Adds DD_TRACE_HTTP_CLIENT_ERROR_STATUSES, LLM Observability experiment concurrency, and agentless log submission for Bunyan and Pino in test optimization.
Feature Flags gain agentless exposure routes with a fallback, and AppSec now reports duration metrics for WAF and RASP. Also includes LLM Observability dataset record tags and a performance improvement to carrier extraction overhead.
Adds direct EVP proxy intake routes, feature flag exposure route fallbacks, and agentless exposure route selection, plus LLM observability dataset record tags. Fixes include safer stringification of dynamic instrumentation template values and preserving quarantine across EFD retries.


