Remember, the idea is to use as much as possible from the kit to build a 40m QRSS transceiver.
First consideration, phase 1 completes the keying and rx/tx-switching. We need that, phase 1 should be built w/o any alterations.
Phase 2: this is the VFO & buffer, well, we don't want a VFO, we want an (TC)XO instead, hence, I propose to postpone any construction to a later stage. If you really want to do some soldering in that stage, that's seems reasonable to do: Q18, C33, C83, C84, R55-R59, L16. This is the last buffer stage with a low-pass. At the other C33 we will couple in the XO.
Phase 3: TX mixer. I believe, the transmit mixer should be postponed. Really. Let's first the the receiver operational and take care about the transmitter later.
Phase 4: RX front-end and filter. The front-end we really need, I suggest to build everything but the filter. Since the bandwidth we are interested in is just 100Hz, the idea is not to use all four crystals for the filter, but just a single one. Here, some experimentation from my side is still required.
Phase 5: BFO, RX mixer, audio. BFO and RX mixer we obviously need... What audio is concerned, the preamp-stage seem reasonable to build, up to C12. R6 could be replaced by a trimmer and the a.f. final skipped totally, finally we want to feed the signal to the computer.
My preliminary opinion on the AGC is not to include it. It could be turned into a manual gain control (using R6 feeding bias to Q7's base) or made switchable.
UPDATE: All those considerations brought me to a point where so few is left of the original kit, that it would be the best to start from scratch, w/o a kit, with a given bonus.