# slogbox

A `slog.Handler` that keeps the last N log records in a fixed-size circular buffer. Zero external dependencies — just the Go standard library.

**Two main use cases:**

1. **Debug endpoint** — expose recent logs via an HTTP handler at `/debug/logs`. Useful for health checks or admin panels where you want to see what happened recently without digging through log aggregation.

2. **Black box recorder** — inspired by `runtime/trace.FlightRecorder`. Buffer quietly collects logs, and when an error-level record arrives, it auto-flushes the entire buffer (including the context leading up to the error) to a destination handler. You get the "what happened before the crash" context for free.

**Basic usage:**

```go
rec := slogbox.New(500, nil)
logger := slog.New(rec)
slog.SetDefault(logger)

// expose at /debug/logs
http.Handle("GET /debug/logs", slogbox.HTTPHandler(rec, nil))
```

**Black box pattern:**

```go
rec := slogbox.New(500, &slogbox.Options{
    FlushOn: slog.LevelError,              // trigger on errors
    FlushTo: slog.NewJSONHandler(os.Stderr, nil), // dump context here
    MaxAge:  5 * time.Minute,              // only recent records
})
logger := slog.New(rec)

logger.Info("request started", "path", "/api/users")
logger.Info("db query", "rows", 42)
logger.Error("query failed", "err", err)  // flushes all buffered context
```

**Performance:** 150 ns/op and 1 alloc per `Handle` call. With `GOEXPERIMENT=jsonv2`, `WriteTo` uses streaming JSON — 4x less memory, 54x fewer allocs at 100 records, scaling to 27x faster at 10K records.

**Config options:** `Level` (min level to store), `FlushOn` (level that triggers flush), `FlushTo` (destination handler), `MaxAge` (exclude stale records from reads).

**Repo:** https://github.com/alexrios/slogbox — 19 stars, GPL-3.0 license.