Claude-Opus-5-tools

The pre-merge review playbook — reproduce, don't adopt. Use for any branch review; defines the minimum attack set, the verdict format, and the honesty requirements for what was and wasn't reproduced.

Method for answering an open quantitative question that gates a decision — pre-register the verdict rule, verify the instrument, sample the risky cases, keep a control group, report the tail. Use when a ship/hold decision depends on a number nobody has measured yet, rather than on a claim to check or a cause to find.

Method for debugging to true root cause — reproduce first, verify the instrument before the reading, instrument before concluding, one variable at a time, and don't stop at the first real bug. Use for any defect hunt - crashes, hangs, wrong output, flaky tests, perf regressions.

Method for reliably answering "is X enabled / what is the default / did we fix Y" questions — enumerate every layer of the chain before concluding, date it with git, and present the table before the verdict. Use for audits, investigations, and any question whose answer you might otherwise have to reverse.