# Systematic Debugging

> Use when a verification gate fails and the next fix needs reproducible diagnosis instead of speculative patching.

- Skill: `dyai2025/systematic-debugging` (Agent Skill)
- Install (CLI): `npx skillmds@latest add dyai2025/systematic-debugging`
- Raw SKILL.md: https://api.skillmd.com/api/skills/dyai2025/systematic-debugging/raw
- Safety review: pending
- Works with: Claude Code, Claude.ai, OpenAI Codex
- Category: Coding & Dev Tools
- Author: DYAI2025 (https://skillmd.com/u/dyai2025)
- Updated: 2026-09-17
- Page: https://skillmd.com/skills/dyai2025/systematic-debugging

---


# Systematic Debugging

## Loop
1. Reproduce the failure with the smallest exact command.
2. Capture expected vs actual behavior.
3. Form one hypothesis at a time.
4. Add instrumentation or a focused test to falsify it.
5. Apply the smallest fix, then rerun the failing command and the relevant regression suite.

Never batch unrelated fixes in one debugging loop.