debuglisted

# Systematic debugging The failure mode this skill exists to prevent: patching the symptom before understanding the cause. You may not write a fix until phase 3 is complete. ## Phase 1 - Reproduce Turn the symptom into a command that fails deterministically: a failing test, a script, a curl. If you cannot reproduce it, that IS the investigation - gather logs/inputs until you can. Record the exact command; it becomes the regression test later. ## Phase 2 - Isolate Shrink the search space with evidence, not intuition: - Bisect: inputs (minimize the failing case), code path (git bisect if a regression), data (which record breaks it). - Instrument: add targeted logging/asserts at the boundaries you suspect; read the actual values, don't guess them. - One variable at a time; write down what each probe ruled out. ## Phase 3 - Root cause State the root cause in one sentence: "X happens because Y." Then try to disprove it: name one alternative explanation and show why the evidence excludes it. If you can't exclude it, return to phase 2. A cause you can't defend against one alternative is a hypothesis, not a diagnosis. ## Phase 4 - Fix + regression-proof - Write the regression test FIRST (the phase-1 reproduction, minimized). It must fail before the fix and pass after. - Fix the root cause, not the site of the symptom. If they differ, say so explicitly. - Run `./check.sh` (full). Remove the phase-2 instrumentation. - If the bug corresponds to a finding in `.ritual/findings/`, men