Skills for Testing Distributed Systems (github.com)

A new set of skills for AI coding agents has been introduced to enhance the testing of distributed systems. Developed to work seamlessly with various AI platforms like Claude Code, Codex, and Copilot CLI, these skills comprise two main functionalities: one for designing a comprehensive test plan and the other for executing it. The test plan is structured similarly to a Jepsen analysis, focusing on claims, hypotheses, and scenarios designed to challenge specific claims under potential fault conditions. This structured approach not only ensures thorough testing but also delivers an explicit argument for coverage adequacy and lists any untested areas. The significance of this development lies in its ability to address the inherent complexities of distributed and stateful systems, which traditional integration tests often overlook. By facilitating a claim-driven testing approach—where every scenario is designed to validate a specific promise to the user—the skills aim to reduce the occurrence of critical failures caused by overlooked issues such as network partitions or concurrency faults. The execution skill also reuses existing tools and metrics from the system under test, ensuring a streamlined testing process that emphasizes accountability through detailed reporting.