We are working primarily in Linux/terminal/SSH sessions/AWS contexts. We tried the Anthropic CLI, but the invocation rate limits made it cost‑prohibitive. Ended up building an in-house client to interoperate with inference providers with public models.

Interested to hear from developers working in different environments - all opinions are welcome. Thanks!