Describe the feature or problem you'd like to solve
Problem: the latest rate-limit chaos
Proposed solution
Besides the discussion here:
#2827
Some more things:
- => extended to per model family rate limit display (e.g. gpt/claude)
- => the agents should be able to access the actual rate limits (and see the notifications), so they can adapt.
- => a prompt or subagent call can contain a rate limit (we can set a limit ourselves, for each call)
Those are must-have constructs in order to be able to control workflow behavior decently.
Example prompts or workflows
Do this and that.
/token-limit: n
Additional context
META: (this form) this really needs to be a multiline input:
Describe the feature or problem you'd like to solve
Problem: the latest rate-limit chaos
Proposed solution
Besides the discussion here:
#2827
Some more things:
Those are must-have constructs in order to be able to control workflow behavior decently.
Example prompts or workflows
Do this and that.
/token-limit: n
Additional context
META: (this form) this really needs to be a multiline input: