There is no xtal-driver within the CY8CLED04G01-chip, so you have to provide your own clock source with appropiate voltage-level. What is the reason not to use the internal oscillators? Are you running a RTC? As long as you depend on the IMO the stability (within supply and temperature) would be alright even for UARTs ets.
The main reason was UART.
I found a AN that describes temperature compenation for IMO (AN 55973).
That also assured me that IMO stability is sufficient for UART.
Another reason is RTC as an option in the design.
So I'll think it over again and, if needed, provide a seperate crystal oscilator.
Rather late I came over this post again and I've got an idea:
When you need the option for an RTC and your PSoCs is driving LEDs (Ambient light??) there will be a master controlling the LED-PSoC. This can be used to synchronize the master's RTClock (with x-tal) to the slave, saving this way some hardware. The only point to watch-out for is: Not to use the internal low-power oscillator (ILO) because its specs are rather worse and absolutely unusable for an RTC.