What's New in TraceKit
Stay up to date with the latest features, improvements, and updates
AI-Powered Operations - Root Cause Analysis, NL Alerts & Runbooks
Released on March 17, 2026
When something breaks, the system automatically identifies the root cause across service boundaries and explains it in plain language. Create alert rules by describing them in natural English, and every fired alert comes with an AI-generated investigation runbook with evidence chains, deep links, and remediation suggestions.
Key Highlights
Cross-Service Root Cause Analysis
When an anomaly is detected, the system automatically traverses the service dependency graph, identifies the probable root cause service and span, and produces a natural language explanation with confidence scoring and evidence chain.
Natural Language Alert Creation
Describe alert rules in plain English via the AI copilot — "alert me if auth-service p99 exceeds 500ms" — and the system generates a real threshold or anomaly alert rule with a preview card for confirmation before activation.
AI Investigation Runbooks
Every fired alert auto-generates a contextual runbook with structured investigation steps, deep links to relevant traces and anomalies, similar past incidents, and suggested remediation — clearly labeled as suggestions, not authoritative fixes.
RCA Evidence Chain
Root cause analysis includes a step-by-step evidence chain showing how a fault propagated through your services, with each step labeled as root cause, propagation, or symptom and linked to the actual trace spans.
Enriched Alert Notifications
Alert emails, Slack messages, and webhooks are automatically enriched with runbook summaries and RCA results. Webhook payloads include structured JSON fields for programmatic consumption.
Feedback Loop
Thumbs up/down feedback on RCA results helps improve analysis quality over time. Runbooks are cached per alert rule and service to deliver instant results on repeated incidents.
LLM Observability & AI Copilot - See Inside Your AI
Released on March 16, 2026
Full-stack LLM observability with auto-instrumentation for OpenAI and Anthropic across Node.js, Python, and Go SDKs. Track every LLM call with cost, token usage, and latency breakdowns. Plus, an AI copilot that answers questions about your system using real data — search traces, check anomalies, and investigate issues through natural conversation.
Key Highlights
AI Copilot
Ask questions about your system in plain English. The copilot searches traces, checks anomalies, inspects service metrics, and browses LLM call data — all through a conversational interface accessible from any page.
LLM Auto-Instrumentation
Drop-in SDK support for OpenAI and Anthropic in Node.js, Python, and Go. Every LLM call is automatically captured with model, tokens, cost, duration, and prompt/completion metadata — no code changes required.
LLM Performance Dashboard
Dedicated dashboard for LLM operations with cost breakdown by model and service, token usage trends, latency percentiles, and error rate tracking. See exactly where your AI spend is going.
LLM Cost & Anomaly Alerts
Automatic detection of LLM cost spikes, token surges, and unexpected model changes. Get alerted when spending deviates from baselines before costs spiral.
AI Gateway with Token Budgets
All AI features route through a metered gateway with per-organization token budgets. When budget is exhausted, non-critical AI features gracefully degrade instead of failing.
Distributed Trace Correlation
LLM calls appear as spans in your distributed traces, so you can see AI latency in context alongside database queries, HTTP calls, and other operations.
Code Monitoring v2 - Live Breakpoints, PII Scrubbing & Kill Switch
Released on March 7, 2026
A complete evolution of Code Monitoring across all 8 backend SDKs. Set breakpoints in production code and stream variable snapshots in real-time via SSE. Sensitive data is automatically scrubbed before it leaves the process, and a per-service kill switch lets you disable monitoring instantly.
Key Highlights
Automatic PII Scrubbing
All snapshots are scanned for sensitive fields (password, token, secret, api_key, credentials) before leaving the process. Values are replaced with [REDACTED] using boundary-aware pattern matching across all 8 SDKs.
Real-Time SSE Streaming
Breakpoint events stream from server to SDK over Server-Sent Events. New breakpoints activate in seconds without waiting for the next poll cycle, and the dashboard updates live as snapshots arrive.
Per-Service Kill Switch
Instantly disable code monitoring for any service from the dashboard. The kill switch propagates via SSE in real-time — SDKs stop capturing snapshots and slow polling to 60-second intervals until re-enabled.
Circuit Breaker Protection
Built-in circuit breaker prevents snapshot storms. If a breakpoint fires too rapidly, the SDK automatically throttles capture to protect application performance.
All 8 Backend SDKs
Full feature parity across Go, Node.js, Python, PHP, Laravel, Ruby, Java, and .NET. Every SDK supports live breakpoints, PII scrubbing, SSE streaming, kill switch, and circuit breaker.
Session Replay - See Exactly What Your Users Experienced
Released on March 2, 2026
Record and replay user sessions with privacy-first DOM capture. When an error occurs, jump straight to the replay and see what the user was doing, with synchronized network requests, console logs, and direct links to backend distributed traces.
Key Highlights
Pixel-Perfect DOM Recording
Captures full page snapshots and incremental DOM mutations using rrweb. See exactly what the user saw, including dynamic content, scrolling, and interactions.
Privacy-First by Default
All text, input values, and media are masked by default. Selectively unmask non-sensitive elements with CSS selectors. Your users' data stays private without any configuration.
Error-Linked Replays
Every error event carries a replay_id. Click "View Replay" on any error detail page to jump directly to the session, automatically seeked to 5 seconds before the error.
Network & Console Tabs
Synchronized sidebar shows fetch/XHR requests with method, URL, status, and duration, plus console output lined up with the playback timeline. Click "View Trace" on any request to see the backend trace.
Zero-Jank Compression
Recording data is compressed off the main thread in a Web Worker. Chunks upload every 30 seconds, limiting data loss to at most 30 seconds on tab close.
Smart Sampling Modes
Configure session sampling (capture X% of all sessions) and error sampling (capture 60-second buffer around errors) independently to control costs without losing critical replays.
Release Tracking & Source Maps - Ship With Confidence
Released on March 2, 2026
Track every release with crash-free session rates, regression detection, and deploy history. Plus, upload source maps to resolve minified JavaScript stack traces back to your original source code, automatically from your CI/CD pipeline.
Key Highlights
Release Health Dashboard
See crash-free session rates, new issue counts, and regressed issue counts per release at a glance. Sparkline charts show trends over time with environment filtering.
Automatic Regression Detection
Issues you've resolved that reappear in a newer release are automatically flagged as regressions so you catch re-introduced bugs before your users do.
Source Map Symbolication
Upload source maps via CLI or build plugins, and minified stack traces are automatically resolved to original file names, line numbers, and function names.
Webpack & Vite Build Plugins
Drop-in @tracekit/webpack-plugin and @tracekit/vite-plugin auto-inject debug IDs and upload source maps as part of your build process with zero manual steps.
Auto-Register from OTel Data
Server-side releases are automatically registered from service.version in existing OpenTelemetry span data without requiring the CLI for backend services.
CLI for CI/CD Integration
Create releases, register deploys, and upload source maps from your CI/CD pipeline with tracekit-cli. Available via Homebrew, direct download, or install script.
Browser SDK - Full-Stack Observability From Browser to Backend
Released on March 2, 2026
Introducing @tracekit/browser - a zero-dependency JavaScript SDK that captures frontend errors, breadcrumbs, and distributed trace context. Connect browser errors directly to backend traces with W3C traceparent propagation, plus framework-specific wrappers for React, Vue, Angular, Next.js, and Nuxt.
Key Highlights
Zero-Config Error Capture
Auto-captures uncaught JavaScript errors and unhandled promise rejections with zero config after calling init(). Includes stack traces, breadcrumbs, and user context.
Browser-to-Backend Trace Correlation
Automatically injects W3C traceparent headers on fetch and XHR requests, connecting browser spans to backend distributed traces in a unified waterfall view.
Automatic Breadcrumbs
Records console calls, network requests, DOM clicks, and navigation changes as breadcrumbs, giving you the full story of what the user did before an error occurred.
Framework Wrappers for React, Vue, Angular, Next.js & Nuxt
Native error boundaries and navigation breadcrumbs for every major framework. React ErrorBoundary, Vue errorHandler, Angular DI ErrorHandler, Next.js multi-runtime, and Nuxt composables.
Offline Buffering & Reliable Delivery
Events are buffered when the user goes offline and retried with exponential backoff when connectivity resumes. Uses sendBeacon on page unload for reliable delivery.
Lightweight & Zero Dependencies
Just 6KB gzipped with zero runtime dependencies. Triple-format builds (ESM, CJS, IIFE) work everywhere, from modern bundlers to plain script tags.
Custom Metrics - Track What Matters Most to Your Business
Released on February 2, 2026
Go beyond traces with custom metrics that capture the pulse of your application. Track counters, gauges, and histograms to measure business KPIs, system health, and performance trends—all without leaving TraceKit.
Key Highlights
Measure Business Impact, Not Just Technical Data
Track revenue per transaction, orders processed, user signups, or any custom KPI. Transform raw application data into actionable business insights.
Real-Time Visibility Across All Services
See metric values update in real-time across your entire stack. Instantly identify which service is dropping, spiking, or behaving unexpectedly.
Three Metric Types for Every Use Case
Counters for tracking totals (requests, errors), Gauges for current values (queue depth, connections), and Histograms for distributions (latency percentiles, response sizes).
Beautiful Metrics Explorer
Browse, search, and analyze all your metrics in one place. View time series charts, statistical summaries, and tag breakdowns without writing queries.
Add to Any Dashboard in One Click
Pin important metrics directly to your dashboards. Build unified views combining traces, traffic data, and custom metrics side by side.
Drop-In SDK Support for Every Language
Simple, consistent API across Go, Python, Node.js, PHP, Laravel, and Java. Add metrics to your app in under 5 minutes with zero infrastructure changes.
DDoS & Traffic Protection - Real-Time Attack Monitoring
Released on January 31, 2026
Introducing comprehensive DDoS and traffic attack protection monitoring. Automatically detect volumetric attacks, rate limit violations, and suspicious traffic patterns in real-time with intelligent severity classification and instant notifications.
Key Highlights
Real-Time Attack Detection
Automatically identifies volumetric attacks, rate limit violations, rapid requests, and suspicious patterns across all your services and endpoints
Intelligent Severity Classification
Smart severity levels (Critical, High, Medium, Low) based on request volume, time window, and attack patterns - focus on what matters most
Comprehensive Traffic Dashboard
Dedicated /traffic page with real-time statistics, top attacking IPs, attack pattern breakdowns, and filterable event timeline
Instant Email Notifications
Automatic alerts sent to organization owners when attacks are detected - stay informed even when you're away from the dashboard
Deep Attack Analysis
View client IP, user agent, HTTP method, status code distribution, geographic origin, and endpoint targeting for each attack
Event Management Workflow
Mark events as resolved or false positives with notes, filter by status, severity, and attack pattern - complete incident response workflow
Security Insights - Detect Sensitive Data & SQL Injection Risks
Released on January 29, 2026
TraceKit now includes Security Insights to help you identify security-relevant patterns in production traces. Automatically detect sensitive data exposure (passwords, API keys, tokens) in code snapshots and identify potential SQL injection risks in database queries - all without becoming a full security product.
Key Highlights
Sensitive Data Detection
Automatically scans code snapshots for passwords, API keys, JWT tokens, and credit card numbers - redacts them before storage
SQL Injection Pattern Detection
Analyzes database queries for suspicious patterns including UNION injection, time-based attacks, and boolean injection attempts
Security Events Dashboard
New dedicated Security Events page with filtering by severity, status, and event type - includes acknowledge and false positive marking
Security Alert Rules
Configure custom alert rules for security events with email, Slack, and Telegram notifications plus automatic fallback to org admins
Webhook Integration
Security alerts automatically trigger webhooks for high/critical events with full event details and trace links for external integrations
All SDKs Updated
PHP, Laravel, Node.js, Go, and Python SDKs all include security scanning - v1.0.5+ for PHP, v1.2.0+ for Laravel, Go, Python, and v1.3.0+ for Node.js
Local UI - Debug Traces Without Cloud Setup
Released on January 18, 2026
Introducing TraceKit Local UI - a standalone development tool that lets you see traces instantly on your local machine without creating an account or getting an API key. Perfect for rapid development and demos with setup under 60 seconds.
Key Highlights
Zero-Configuration Setup
Install with npm and start seeing traces in under 60 seconds - no account creation, no API keys, no cloud dashboard required
Real-Time Local Dashboard
Beautiful web dashboard at localhost:9999 with WebSocket live updates, trace statistics, and service monitoring
All SDKs Auto-Detect
Node.js, Python, PHP, Laravel, and Go SDKs automatically detect and send to Local UI in development mode - no configuration needed
Smart In-Memory Storage
Stores up to 1000 traces with 1-hour retention, automatic cleanup of old traces, and sub-millisecond ingestion performance
Dual-Send Support
SDKs send traces to both local UI and cloud simultaneously when API key is configured - perfect for testing before deployment
Developer-First Design
Solves the #1 adoption blocker - stay in your flow without context switching to cloud dashboards during active development