Nice.
1) Why are there two 100nf caps I the feedback loop? Isn't one sufficient?
2) If you've going to use 3 chips, may as well make the third chip a quad, rather than dual, op-amp. I know the 4066 only has 4 switches on it, but you can use the additional pair of op-amps for two fixed stages, as Rick Holt did for the Causality phaser.
3) The 40106 has 6 inverter stages, but you've only used 4. That's not wrong, but there is the option of adding a second LFO to sum with the first one, and introducing some aperiodicity, or simply having two different rate presets to select between.