New open-source AI coding skills enable automated, claim-driven testing for distributed and stateful systems by generating structured test plans and detailed findings reports for various AI agents.

Key Points

• The framework provides two primary skills: one for designing test plans based on product claims and another for executing those tests.
• It supports major AI coding agents including Claude Code, Cursor, Copilot CLI, Gemini, and Codex.
• Outputs include a structured Markdown test plan and a findings report featuring a 9-state verdict taxonomy and blame classification.
• The system enforces rigorous testing standards, such as linearizability checking, fault injection, and coverage adequacy arguments.
• It includes a comprehensive technique catalog distilled from industry-standard research on distributed system failures and testing methodologies.

Why it Matters

This framework shifts distributed system testing from informal integration checks to a structured, evidence-based workflow that significantly improves production reliability. By automating the design and execution of complex fault-injection scenarios, it allows developers to identify critical bugs—such as non-deterministic concurrency and crash-recovery issues—that traditional testing often misses.