SDK Catalog
Coverage of official and planned SDKs
This catalog is a decision layer, not only a list. Use it to choose the right rollout path per runtime and team maturity.
Official SDKs
| SDK | Status | Primary Runtime | Best For | Docs |
|---|---|---|---|---|
| JavaScript Browser | GA | Browser | User-facing telemetry and frontend correlation | Open |
| JavaScript Node/Bun | GA | Node.js / Bun | APIs, workers, edge-compatible JS services | Open |
| Python | GA | CPython | FastAPI/Django workers and data services | Open |
| Go | GA | Go 1.21+ | High-throughput services and low-latency APIs | Open |
| Java | GA | JVM 17+ | Enterprise microservices and Spring ecosystems | Open |
| .NET | GA | .NET 8+ | ASP.NET APIs and enterprise backend workloads | Open |
| PHP | GA | PHP 8.2+ | Laravel and traditional web backend stacks | Open |
| Ruby | GA | Ruby 3.2+ | Rails services and background job ecosystems | Open |
Planned / In Validation
| SDK | Status | Notes |
|---|---|---|
| Rust | Beta Planning | Focus on transport safety and tracing ergonomics. |
| Kotlin | Beta Planning | Focus on Ktor and Spring compatibility. |
| Swift | Beta Planning | Focus on iOS telemetry and offline buffering. |
Selection Guide
Choose SDK priority using this order:
- Service criticality (incident cost if it fails).
- Request volume (signal usefulness).
- Team readiness (who can own rollout and support).
- Runtime constraints (cold starts, memory, edge limits).
Shared Conventions Across SDKs
- Auth model is consistent (API key + ingest base URL + tenant/project context).
- Core primitives are aligned (log/trace/error + flush/shutdown).
- Retry/queue behavior is runtime-specific but documented.
- Troubleshooting taxonomy is shared (auth, transport, volume, latency, cardinality).
Rollout Pattern by Runtime
- Start with one service per language.
- Validate data quality gates.
- Expand by topology (sync API -> async workers -> edge/frontend).
- Add release metadata and runtime integration before scaling rollout.
Definition of Done (Per SDK Rollout)
- Instrumentation in place and peer-reviewed.
- Telemetry visible with stable tags.
- Error and trace correlation proven in at least one incident simulation.
- Runbook exists for auth rotation and transport failure recovery.