-Infinity is a (potentially harmful) side effect rather then the actual bug. Your second test, "Mathf.Abs(-0f) = 0" can't test the output because the correct IEEE 754 behavoir is for -0 and 0 to be treated as equal by the comparison operator.

However with bit testing you can determine if a float is 0 or -0 which exposes that Mathf.Abs(-0) is actually returning -0 when it should be returning 0. This may be a Mono framework bug. You can update your project with this struct so it can directly test whether the sign bit is set (negative) or not set (positive).

// using System.Runtime.InteropServices
[StructLayout(LayoutKind.Explicit)]
public struct FloatBits
{
const uint NegativeBit = 1u << 31;

[FieldOffset(0)]
public float Value;

// Shares the same address in memory as Value, allowing its bit to be read and modified
[FieldOffset(0)]
uint Bits;

// Set to true to make a value negative, false to make it positive.
public void SetNegativeBit(bool value)
{
if (value)
{
Bits |= NegativeBit;
}
else
{
Bits &= ~NegativeBit;
}
}

public bool IsNegative()
{
return (Bits & NegativeBit) != 0;
}
}