Multi-agent coordination for prod dev

A developer has been using AI agents to automate product development, focusing on pre-development aspects like architecture and design. They found that while a single model like Opus 5 performs well, using multiple agents from different LLM families, such as Google's Antigravity, OpenAI's GPT Sol, and Anthropic's Opus 5, yields better results. They also use Fable and Astra for critical tasks. The developer built an internal orchestration router but is seeking existing tools that support native CLIs over API keys for multi-agent coordination.

This report details a developer's practical experience using a combination of Google's Antigravity, OpenAI's GPT Sol, and Anthropic's Opus 5, unlike others that focus on single-model applications.