sui

This skill provides a structured framework for diagnosing complex, non-deterministic-seeming failures within the Sui blockchain's simulation testing environment. Designed for engineers and developers working on the Sui codebase, it utilizes the scientific method to isolate root causes in integration tests. By systematically logging system state, observing execution flow, and iteratively testing hypotheses, users can pinpoint issues without introducing functional code changes that might mask the underlying bug.

• Executes cargo simtest with specified test targets and reproduction commands.

• Automates seed searching using seed-search.py to identify failing test cases.

• Maintains a rigorous NOTEBOOK.md audit trail of observations, hypotheses, and experimental results.

• Facilitates clean debugging sessions through standardized logging patterns (info!("CLAUDE: ...")) and structured commit strategies.

• Supports long-running test execution with full output redirection and diagnostic persistence.

• Requires a reproduction command or test name as the primary input.

• Strictly prohibits functional changes to the codebase; debugging must rely solely on non-intrusive logging.

• Adheres to a strict cleanup protocol: all debugging commits, logs, and NOTEBOOK.md files must be purged prior to final PR submission.

• Works best with deterministic simulation tests; if a reproduction stops failing, the system assumes a functional change occurred.

• The tool is specifically tuned for the MystenLabs Sui repository architecture, including standard Rust logging and simtest infrastructure.