SDK Catalog

Choose the right SDK rollout path based on what you are trying to instrument first.

SDK Catalog

Use this page to decide what to instrument first. Do not start by language preference alone. Start by incident value and operational risk.

Choose by intention

I need backend incident visibility first

Best starting SDKs:

Then follow Instrument your first service.

I need frontend to backend correlation

Start with:

I have a multi-stack project (e.g. React + Supabase)

Obtrace auto-detects all stacks. Instrument each one:

I need platform-specific rollout guidance

Start with:

Official SDKs (GA)

SDKPrimary RuntimeBest first use
JavaScript BrowserBrowserUser-facing telemetry and frontend correlation
JavaScript Node/BunNode.js / BunAPIs, workers, edge-compatible JS services
PythonCPythonFastAPI, Django, workers, data services
GoGo 1.21+Low-latency APIs and high-throughput services
JavaJVM 17+Spring and enterprise microservices
.NET.NET 8+ASP.NET APIs and enterprise backends
PHPPHP 8.2+Laravel and traditional web backends
RubyRuby 3.2+Rails and background jobs

Platform & Edge SDKs (GA)

SDKRuntimeBest first use
Next.jsNext.js 13+Fullstack apps (server + client in one setup)
SupabaseManaged Agent (OAuth)Full Supabase project monitoring (DB, Edge Fn, Auth, Storage, Realtime)
VercelVercel FunctionsVercel serverless functions
CloudflareCloudflare WorkersEdge compute
RailwayAny (Node default)Railway-hosted services

Agents

AgentRuntimeBest first use
Database AgentDockerDatabase monitoring (Postgres, MySQL, MongoDB, Redis, etc.)
Infrastructure AgentDockerHost/container metrics (CPU, memory, disk, network)
Kubernetes (Zero)K8sZero-code cluster-wide auto-instrumentation
MCP ServerGo / DockerAI assistant integration (Claude, Cursor)

Coming Soon (Not Yet Available)

These SDKs are planned but have not been released. No installation instructions are available.

SDKTarget RuntimeStatus
KotlinJVM (Ktor, Spring Boot)Planned
RustRust asyncPlanned
SwiftiOSPlanned

For Kotlin projects, use the Java SDK which works with any JVM language.

Decision rules

  1. Start with the service that hurts most when it fails.
  2. Prefer one production-relevant service over many partial rollouts.
  3. Add release metadata before expanding.
  4. Move to runtime and security guidance after basic telemetry is healthy.