The lowest-hanging fruit here, which you may already have picked, is to spend your low-capacity time on tasks which improve the quality or quantity of your high-capacity time. It's hard to guess what tasks would even qualify as low-capacity for you -- you on a bad day might have an easy time with some particular thing that I would struggle with on a good day, and vice versa. For instance, it might be high-capacity to plan a menu and order groceries, but low-capacity to wash and chop all the vegetables you'll need for a weak of nutritious meals. It might be high-capacity to triage your closet and decide where to store each type of item, but low-capacity to fold a load of clean laundry and place each item in the spot with the corresponding label.

Budget some of your high-capacity time toward reducing the capacity required for inevitable tasks, so that you can offload them to lower-capacity times and free up more high-capacity time for the tasks you find most valuable. For instance, I labeled stuff around the house (big plates go here, small plates go there, bowls go there, this switch is for the outside lights and that switch next to it is for the inside ones) to benefit guests, but I've been amazed by how it reduces cognitive load for me as well. Store metadata about your life outside your head -- I find it helpful to try to treat my future self with at least the courtesy I would show to strangers. I try to code as if an intern will be attempting to understand it; I try to manage my physical workspace as neatly as I would want a public makerspace to be organized.

When you ask for "low-cognitive-workload altruistic tasks", how are you differentiating "altruistic tasks" from others? Is it altruistic work to feed, house, clothe, and generally facilitate the focus and productivity of a person who does "high-cognitive-workload altruistic tasks"? I would argue that any task with which the obviously altruistic work couldn't happen is itself altruistic. You are your own support system; give yourself credit for all the roles you occupy that make your "more important" work possible.

One option is writing or editing questions and answers for Stampy, the AI alignment FAQ Rob Miles's community is building.

We're building the reader UI (very early prototype, missing search, automatically adding new questions as you click, and prettiness) and bot interface.

I don't have anything shovel-ready for this, but I propose as a general pattern data gathering. By this I don't mean doing research but just actual yeoman's tasks of finding the data, collating it, and putting it in suitable format for easy access.

An example of a case where this lead to something interesting when the researcher themselves did it is Mandelbrot's Fractal Markets Hypothesis, where Mandelbrot personally collated huge quantities of market data to notice the pattern. There are also a few examples in 12 Things I Learned Studying Nature's Laws.