The tool is still relatively new, so context window management and large codebase
navigation likely need more polish
No mention of agentic loop controls (e.g., approval modes, auto-apply guardrails) trust and safety mechanisms matter a lot for an agent that writes code
Terminal-first UX limits accessibility; there's no IDE extension path mentioned, which narrows the audience
"Model-agnostic" is only as strong as the abstraction layer, if switching models silently degrades quality (e.g., tool-use inconsistencies), users may not notice until it's a problem