Do note, that it may be sadly necessary to reorder the order in which information is presented, to be able to add in a CTA.
But also, adding as much as CTA as possible is preferred, as tinkering with the code reinforces the information presented to them, by confirming with experience, that the information presented to them is indeed correct.
I don't know the terminology to use here, but I hope that my intent is understood.

With that in mind, here are some of my observations that immediately jump out to me:

In part 2, Getting started, I think a reordering may be necessary to make the custom logo part more engaging. It's begging to have some extremely easy engagement project, like typing it out by hand, swapping the logo every 60 frames between two animation frames, playing around with palettes, or using DB with binary to make them bruteforce the tile format by experimentation.
In part 3, Story Screen, there are even multiple tasks that could be done!
- Hold B to increase text rendering speed to 1 char per frame
- Press Select to skip the intro entirely
- (optional, harder difficulty) Press A or something to type all text out at once while it's typing, and wait for A to be released before returning from the function to avoid detecting A press as going to the next chapter.
In part 3, Heads-Up Interface, there are a few that are possible
- Change the size of the HUD from 8 to 16 tall
- (medium-hard) Move the HUD from the top to the bottom
In part 3, section doesn't really matter, it is possible to add custom metasprites with overlapping graphics, for extra graphical fidelity
(very difficult) In part 3, don't know which section this could fit into, add animation to the bullets by keeping track of the amount of frames they have been alive for, and changing the graphics frame every 2^N frames or so, based on how long they have been alive for. Alternatively, use two counters, one for current frame, and other for a down-counter to count how many frames until having to update the animation.
(medium-difficult) In part 3, collision detection section, reduce the size of the collision detection, and also add support for offsetting the collision box compared to the position inputs, so the collision detection box can be reduced to what "feels right" as opposed to what the graphics tell us the collision box should be.
In part 3, after finishing the tutorial, add random numbers, and add in X velocity to bullets, so they don't just fire directly downwards, but come in at a slight angle for each bullet, adding some depth to the game.

There might be definitely more that I may have accidentally missed, but things are a bit too incomplete at the moment to be able to easily point at something with high confidence and go "this DEFINITELY needs to be tinkered with!".

Uh oh!