Show HN: Claude Code skills and hooks that prove a fix's test fails without it (github.com)

A new plugin for Claude Code, named "agent-workflow," has been introduced, designed for engineers to enhance the process of debugging and code management. This tool enables a supervised workflow where tasks can be divided among various models, with a mechanism to elevate the quality of assistance based on previous failures. Significantly, it includes features like automated handoff notes, failure-proof testing, and skill collision audits, all functions that streamline the software development process by ensuring that fixes are validated against specific test cases that demonstrate a failure without the proposed solution. The implications for the AI/ML community are profound, as this plugin represents a shift toward more collaborative and efficient coding environments, wherein AI agents can work together seamlessly. Specifically, the integration of tests that fail without a fix serves to improve software reliability, while the separation of tasks into tiered agents enhances resource allocation by utilizing cheaper models first. The technical architecture, including the new skills and hooks, not only fosters better collaboration but also provides a robust framework for continuous integration, validating each step before moving onto the next, thus contributing to more resilient code development practices.