Open Issues Need Help
View All on GitHub The CLI cannot use a real tokenizer, only the built-in estimate about 2 hours ago
enhancement help wanted
Measure and cut the token cost of your LLM and agent message payloads. Zero model calls.
JavaScript
Numeric flags accept anything and fail silently as NaN about 2 hours ago
bug good first issue
Measure and cut the token cost of your LLM and agent message payloads. Zero model calls.
JavaScript
compact --max re-tokenises the whole payload on every dropped message about 2 hours ago
bug good first issue
Measure and cut the token cost of your LLM and agent message payloads. Zero model calls.
JavaScript