prompt logo

prompt

Free

A disciplined diagnosis loop for hard bugs and performance regressions.

FreeFree tier
Inputs: textOutputs: text
Type
Open Source

About prompt

A structured prompt for debugging, guiding users through a disciplined diagnosis loop for hard bugs and performance regressions. It emphasizes building a fast, deterministic feedback loop as the key skill, and provides multiple techniques to construct such loops, including failing tests, curl/HTTP scripts, CLI invocations, headless browser scripts, replay traces, throwaway harnesses, property/fuzz loops, bisection harnesses, differential loops, and HITL bash scripts. The prompt covers phases from building a feedback loop to iteratively improving it, then hypothesizing, instrumenting, fixing, and regression testing.

Key Features

Systematic debugging methodology
Emphasis on fast, deterministic feedback loop
Multiple techniques for constructing feedback loops (failing test, curl, CLI, headless browser, replay trace, throwaway harness, property/fuzz, bisection, differential, HITL)
Iterative improvement of the loop for speed, sharpness, and determinism
Phases: reproduce, minimise, hypothesise, instrument, fix, regression-test

Pros & Cons

Pros
  • Provides a structured methodology for debugging complex issues
  • Encourages building fast, deterministic feedback loops to accelerate debugging
  • Covers a wide range of debugging techniques adaptable to different environments
  • Emphasizes iteration and continuous improvement of the debugging process
Cons
  • May be overly rigid for simple or trivial bugs
  • Requires technical expertise to implement many of the suggested techniques
  • Some techniques (e.g., property/fuzz loops, bisection) can be time-consuming to set up

Best For

Debugging hard bugsDebugging performance regressionsSystematic problem diagnosis in software development

FAQ

What is the purpose of this prompt?
To guide users through a disciplined diagnosis loop for hard bugs and performance regressions, with steps like reproduce, minimise, hypothesise, instrument, fix, and regression-test.
What is the key skill emphasized in the prompt?
Building a fast, deterministic, agent-runnable pass/fail signal for the bug. The prompt states that this skill is the most important and everything else is mechanical.
What techniques are suggested for building a feedback loop?
Suggested techniques include failing test, curl/HTTP script, CLI invocation, headless browser script, replaying a captured trace, throwaway harness, property/fuzz loop, bisection harness, differential loop, and HITL bash script.