Format Kubernetes byte quantities for humans.

• Zero dependencies (just Go stdlib)
• Parses Kubernetes resource.Quantity formats
• Designed for readability, not precision

Kubernetes resource.Quantity only outputs human-readable format (like 170Mi, 12Gi) when the value is exactly divisible into binary units:

1073741824 bytes  → "1Gi"       (exact)
178257920 bytes   → "170Mi"     (exact)
178255984 bytes   → "178255984" (arbitrary - not human readable)

Additionally, resource.Quantity preserves the scale at which values were originally set. A value from df output (in KiB) stays in KiB even when it would be more readable as GiB:

"spaceAvailable": "12075408Ki"  →  hard to read (actually ~11.5 GiB)

Many operators store byte values as raw integers or strings in status fields. This tool converts those values in-place to the most readable unit.

go install github.com/fatih/kfmt@latest

Convert a single value:

$ kfmt 178255984
170MiB

$ kfmt 12075408Ki
11.5GiB

Convert fields in JSON:

$ echo '{"usedBytes": "178255984", "spaceAvailable": "12075408Ki"}' | kfmt --json-fields "usedBytes,spaceAvailable"
{"usedBytes": "170MiB", "spaceAvailable": "11.5GiB"}

Real-world example with kubectl, without and with kfmt:

$ kubectl get cr -o json | jq '.items[].status.storage'
{
  "currentSize": "12884901888",
  "usedBytes": "178255984",
  "spaceAvailable": "12365217792"
}

$ kubectl get cr -o json | jq '.items[].status.storage' | kfmt --json-fields "currentSize,usedBytes,spaceAvailable"
{
  "currentSize": "12.0GiB",
  "usedBytes": "170MiB",
  "spaceAvailable": "11.5GiB"
}

Decimal values with suffixes are supported: 1.5Gi → 1.50GiB

The output is rounded to fit the most readable unit:

• Values ≥100 show no decimal places (100MiB)
• Values ≥10 show 1 decimal place (11.5GiB)
• Values <10 show 2 decimal places (1.50KiB)

Note: Converting between units may lose precision. For example:

• 12075408Ki (12,365,217,792 bytes) displays as 11.5GiB (actual: 11.5146... GiB)
• Decimal SI units (K, M, G) are converted to binary IEC units (KiB, MiB, GiB), which are ~2.4% smaller

This tool is designed for human readability, not precision. Use the original values for calculations.

This tool is for byte/storage values. The following Kubernetes quantity features are intentionally not supported:

• Milli (m), micro (u), nano (n) suffixes — not meaningful for bytes
• Negative values — bytes can't be negative
• Full arbitrary precision — we accept rounding for readability