Guide debugging from observed symptoms to one testable hypothesis per cycle.
npx skills add https://github.com/zeikar/hyperclaude --skill hyper-debug
Please help me install this Agent Skill. Skill: hyper-debug Source: https://github.com/zeikar/hyperclaude/tree/main/skills/hyper-debug Command: npx skills add https://github.com/zeikar/hyperclaude --skill hyper-debug
hyper-debug prevents inefficient guess-and-check debugging by enforcing a disciplined loop from reproduction to hypothesis testing, then to regression protection before any fix.
Invoke hyper-debug when you hit a failing test or unexpected behavior so you can reproduce it, form one hypothesis, run a targeted experiment, and only then write a failing regression test and apply the fix.
High-intent search queries and answers about installing and using this skill.
Hypothesis-driven debugging is a systematic method that turns observed symptoms into one testable hypothesis at a time. It reduces inefficient guess-and-check by reproducing the smallest failing case, stating expected vs actual outcomes, and validating the cause before fixing.
To find the root cause of a test failure, reproduce the smallest failing case, form one testable cause hypothesis, and run a targeted experiment. You predict evidence before executing, confirming the exact cause only after the experiment validates your prediction.
You should write a regression test only after confirming the exact cause of the bug. This ensures the failing test captures the confirmed behavior, protecting against future regressions before you apply the actual fix to the codebase.
Yes, this approach applies to behavior regressions, but it requires reproducing the smallest failing case. If the regression or exception cannot be reliably reproduced to state expected vs actual outcomes, hypothesis testing will not be effective.
The best way to stop trial-and-error debugging is to enforce experiment-first reasoning. By forcing clear proof criteria before running anything, you validate one testable hypothesis against predicted evidence, eliminating wasted iterations and thrash.
h-4vokDiagnose Playwright smoke failures and runtime-visible regressions in diagram-tours.
tomgunReproduce bugs with regression tests and apply minimal root-cause fixes.
samjhechtDiagnose root causes of bugs using a four-phase investigation workflow.
bryonjacobBreaks down bugs by applying hypothesis-driven analysis to identify root causes.
