A discipline for hard bugs. Skip phases only when explicitly justified.
\nWhen exploring the codebase, use the project's domain glossary to get a clear mental model of the relevant modules, and check ADRs in the area you're touching.
\nThis is the skill. Everything else is mechanical. If you have a fast, deterministic, agent-runnable pass/fail signal for the bug, you will find the cause — bisection, hypothesis-testing, and instrumentation all just consume that signal. If you don't have one, no amount of staring at code will save you.
\nSpend disproportionate effort here. Be aggressive. Be creative. Refuse to give up.
\ngit bisect run it.scripts/hitl-loop.template.sh so the loop is still structured. Captured output feeds back to you.Build the right feedback loop, and the bug is 90% fixed.
\nTreat the loop as a product. Once you have a loop, ask:
\nA 30-second flaky loop is barely better than no loop. A 2-second deterministic loop is a debugging superpower.
\nThe goal is not a clean repro but a higher reproduction rate. Loop the trigger 100×, parallelise, add stress, narrow timing windows, inject sleeps. A 50%-flake bug is debuggable; 1% is not — keep raising the rate until it's debuggable.
\nStop and say so explicitly. List what you tried. Ask the user for: (a) access to whatever environment reproduces it, (b) a captured artifact (HAR file, log dump, core dump, screen recording with timestamps), or (c) permission to add temporary production instrumentation. Do not proceed to hypothesise without a loop.
\nDo not proceed to Phase 2 until you have a loop you believe in.
\nRun the loop. Watch the bug appear.
\nConfirm:
\nDo not proceed until you reproduce the bug.
\nGenerate 3–5 ranked hypotheses before testing any of them. Single-hypothesis generation anchors on the first plausible idea.
\nEach hypothesis must be falsifiable: state the prediction it makes.
\n\n\nFormat: \"If
\n
If you cannot state the prediction, the hypothesis is a vibe — discard or sharpen it.
\nShow the ranked list to the user before testing. They often have domain knowledge that re-ranks instantly (\"we just deployed a change to #3\"), or know hypotheses they've already ruled out. Cheap checkpoint, big time saver. Don't block on it — proceed with your ranking if the user is AFK.
\nEach probe must map to a specific prediction from Phase 3. Change one variable at a time.
\nTool preference:
\nTag every debug log with a unique prefix, e.g. [DEBUG-a4f2]. Cleanup at the end becomes a single grep. Untagged logs survive; tagged logs die.
Perf branch. For performance regressions, logs are usually wrong. Instead: establish a baseline measurement (timing harness, performance.now(), profiler, query plan), then bisect. Measure first, fix second.
Write the regression test before the fix — but only if there is a correct seam for it.
\nA correct seam is one where the test exercises the real bug pattern as it occurs at the call site. If the only available seam is too shallow (single-caller test when the bug needs multiple callers, unit test that can't replicate the chain that triggered the bug), a regression test there gives false confidence.
\nIf no correct seam exists, that itself is the finding. Note it. The codebase architecture is preventing the bug from being locked down. Flag this for the next phase.
\nIf a correct seam exists:
\nRequired before declaring done:
\n[DEBUG-...] instrumentation removed (grep the prefix)Then ask: what would have prevented this bug? If the answer involves architectural change (no good test seam, tangled callers, hidden coupling) hand off to the /improve-codebase-architecture skill with the specifics. Make the recommendation after the fix is in, not before — you have more information now than when you started.