Sailfish captures function-level execution traces, console logs, network requests with full headers and bodies, exceptions with stack traces, and async I/O operations. Everything is correlated by request ID so you can follow a single request across your entire stack.

Yes — distributed tracing is built in. Each service hop is tracked with parent request IDs, so you get a unified timeline showing function spans, logs, and exceptions at every hop. No manual instrumentation needed.

Record a session that reproduces the issue. Then use the exception analysis to see grouped error patterns, drill into any exception for full stack traces with code context, and trace the exact execution path that triggered it — including every function call, argument, and return value.

Compare any two sessions side by side. Sailfish automatically detects regressions — functions that got slower, increased error rates, new N+1 query patterns — and classifies them by severity. You can catch issues before your users do.

We're building toward a world where debugging is a conversation, not a scavenger hunt. AI agents that understand your codebase, your telemetry, and your intent — surfacing exactly what you need before you even know to ask for it.

Create notes and tasks as you work — link them to specific code symbols or line ranges. Queue tasks for AI processing, and they'll be picked up with full context: the code, the telemetry, the specs. It's like having a junior engineer who never loses context.

Sailfish analyzes span patterns to detect growing collections, increasing function durations over time, and unclosed resources. It flags potential leaks with the specific functions involved so you can investigate immediately.

TypeScript and Python today, with auto-detection for framework patterns. The instrumentation is lightweight — typically under 5% overhead — and captures everything without requiring you to add logging or tracing code manually.