Show HN: Jev-pilot – Jev picks Claude Code's effort, model and skill per prompt (github.com)

The release of Jev-pilot, a new plugin for Claude Code, introduces a sophisticated decision-making model designed to optimize prompt responses. By sequentially querying Jev, TypeSafe's decision model, Jev-pilot determines the necessary reasoning effort and selects the appropriate subagent model—ranging from Haiku for low-stakes tasks to Opus for complex challenges—prior to each user interaction. This allows users to maintain the flow of conversation with Claude Code while optimizing resource use, reducing costs for simpler tasks. This development is significant for the AI/ML community as it enhances the efficiency and effectiveness of automated coding tasks, presenting a more versatile approach to task delegation and execution. The plugin tracks decisions in real-time and suggests improvements through a decision ledger, providing insights into task performance. Moreover, it incorporates advice on how best to delegate tasks, enabling a dynamic interaction model that adapts to the complexity of user prompts, leading to better overall coding outcomes. As a demonstration of leveraging AI decision-making, Jev-pilot exemplifies the increasing trend towards creating specialized models that augments user productivity in AI-driven environments.