BUG: How do I know if I'm exceeding the context for the model?

[rathboma]

In claude code it shows you how much context is left before it needs to compact. I don't have any sense of this in the AI shell. How do I know if I'm blowing up the model with too much context?

[azmy60]

@rathboma If we add an indicator that shows the tokens count, what will happen if the tokens hit the limit?

a. Compacting (summarizing) - Best solution, but need to plan it out first
b. Do nothing. But tell the user that we hit the limit.
c. Any ideas?