gemini is not able to view files that exist, and I am not able to tag them (despite not being gitignored/geminiignored)

What happened?

is there some bizarre indexing that has a hard limit per project?

some random percentage of the files in my project are simply not visible to gemini cli. My project directory definitely has less than 10 million files, probably far less than 500k. however, I am unable to tag many of them, and gemini is unable to read them without using bash tools (which obviously works just fine)

every time I gitignore another broad swath of unimportant files, it unlocks another large set of files I am then able to tag. this leads me to think that there is a piece of logic that is explicitly hiding broad fractions of my project (including critical code and documentation)

What did you expect to happen?

for gemini to be able to... read a file at a path unless it is gitignored or geminiignored

I would advise, if you are unable to implement ReadFile to a rough MVP level, to essentially just delete the feature and have gemini use cat/sed etc because these work just fine. I am fine with manually pointing gemini at files without tagging, but it is harmful to tell gemini that files do not exist when they are the crux of a refactor.

Client information

0.31 darwin on gemini ultra, m3 max 32gb

Login information

No response

Anything else we need to know?

if there's a limit can you just disable it with MAX_INT_32 or something? my codebase is not big at all, less than 10gb. or if your concern is that enterprise-scale codebases will not load on, like, low-end mobile devices, can you make it a setting?