The Claude file, documentation, and memories you leave in the project are part of the harness. Most of Claude's forgetfulness can be eliminated with strategic placement of documentation. After you've got that figured out, you'll have to compact the session long before it forgets because it starts getting sloppy and slow at high context. With your guidance it will basically keep its own harness updated for you as the project progresses.
And it also helps to give it better tools than grep and find to explore, because it can waste lots of tokens and pollute the context with the defaults.
I'd suggest CLAUDE.md or README.md (and just write in CLAUDE.md that instructions are in README.md, if they're also supposed to be used by human devs) for scripts and other operative instructions.
Ideally also the PLAN MODE for any long form work within the same session, possibly with sub-agents or dynamic workflows to not pollute the main context too quickly.
It also has a memory mechanism but holy shit is it annoying when it saves meaningless stuff that quickly becomes outdated.
And it also helps to give it better tools than grep and find to explore, because it can waste lots of tokens and pollute the context with the defaults.