ai agents write code with json ed; this is stupid goal: make a new code editor for ai agents that is smarter then this bad idea while enforcing better habits on the ai, specifically for my approved subset of dlang
command and daemon: bspwm has 3 programs, the wm itself, a command interface and the hotkey daemon; consider how this was done in a research document. Likewise there should be 2 programs, 1 that contains continuous state and one that parses the commands. Then a 3rd command pallet for humans to control things
Better workflow: 1. todo lists, with nested subtasks 2. test driven devoplment 3. automatic running of code
subset of dlang: 1. warn that I hate oo when the code contains "private"/"class" 2. the top of all files has a #! and this must compile on each addition 3. enforce unittests on each block of code 4. tabs over spaces
deamon: 1. keeps track of current todo list item, current file, current code block 2. actually edits the files, rolls back instantly on failure and task switch 3. listens for commands, gives responses 4. started in the root folder of the project, all commands are relative to this
commands-ai:
1. appends blocks of code and unittests pairs or unittests to a *_tests.d
2. can print all code block headers that match a regex pattern
3. can replace block of code by regex on the header (must have only one match)
4. can update failing unittests freely( unit tests are either numbered in a test file or paired with a code block)
5. add todo list/nagivate it
6. navigate/read files
commands-human: 1. pause/resume ai (if the commands ai has doesnt return the next attempted edit stalls) 2. add todo list items(with notification to ai) 3. force a todo list item selection 4. unmarking a todo list 5. undeleting a unit test and preventing it from being deleted again 6. select target metrics such as number of tests in a file per todo list item before allowing ai to continue 7. run all tests on demand
Deleted unittests are stored the deamon doesnt close todo list items unless a unit test is associated with proving it, it keeps track of this sub todo list tasks (1 depth not a full tree)