karvifi/os/debug-systematic

Trust tier 3
  • community
Source check pending

Checking that the SKILL.md file is still in the repository.

Scientific debugging — hypothesis, test, observe, refine — not random guessing

Install

npx vskill@latest install karvifi/os/debug-systematic

SKILL.md

SKILL: Debug Systematically

The Scientific Method for Debugging

Step 1: Reproduce the Bug
Can you make it happen again?
- Exact steps to reproduce
- Input data that causes it
- Environment (development? production?)
If you can't reproduce → likely race condition or timing issue
If always reproducible → easier to fix
Step 2: Narrow the Scope
Where could this bug be?
❌ Not "somewhere in the code"
✅ "Happens in authentication flow, specifically password hashing"
Test:
- Does it happen in all browsers? (just one = browser issue)
- Does it happen with all inputs? (just some = input validation)
- Does it happen in staging? (yes = code; no = environment)
Step 3: Form a Hypothesis
"I think the bug is [specific location] because [reason]"
Example:
"I think the password hash comparison is using == instead of constant-time comparison,
allowing timing attacks."
NOT: "I think there's a bug"
YES: "I think the function is comparing hashes using string equality which is vulnerable"
Step 4: Test the Hypothesis
python
# Add instrumentation (logging, breakpoints)
logger.debug("about to compare", {
"input_hash": input_hash,
"stored_hash": stored_hash,
"comparison_function": "=="
})
# Run with test data that exercises the code path
# Example: password "test123" with stored hash "..."
# Observe the output
# Does it match your hypothesis?
Step 5: Refine or Move On
If hypothesis was correct:
→ Fix the code
→ Write a test that would catch this
→ Commit with explanation
If hypothesis was wrong:
→ Go back to Step 2
→ Narrow to a different scope
→ Form new hypothesis

Common Debugging Patterns

Bug pattern: "Works on my machine"
Hypothesis: Environment difference (Python version, dependency version)
Test: pip freeze on both machines, compare
Fix: Pin versions in requirements.txt
Bug pattern: "Only happens sometimes"
Hypothesis: Race condition (concurrent requests)
Test: Load test with 100 concurrent requests
Fix: Add locking or transaction boundaries
Bug pattern: "Just started happening"
Hypothesis: Recent change caused it
Test: git log --oneline | head -20
git diff [last-good-commit]
Fix: Revert the change or fix it properly

Quality checks

  • [ ] Bug is reproducible (exact steps documented)
  • [ ] Scope narrowed (not "somewhere in the code")
  • [ ] Hypothesis is specific (not vague)
  • [ ] Hypothesis tested with evidence
  • [ ] Root cause found and verified
  • [ ] Test written to prevent regression
  • [ ] Fix is minimal (not "rewrite everything")

Security

  • Security scanPassed, 95 of 100Pattern checks on the skill's files.
  • Model reviewPassed

Quality evaluation

  • Task evaluationNot runNo evaluation yet. Run one in Skill Studio.

Source

Popularity

7-day trend score
5

Not refreshed yet.

Works with

Any agent vskill installs to: 56 targets in all. These 39 have a page on this site.

  • Claude Code
  • Cursor
  • GitHub Copilot
  • Windsurf
  • Codex
  • Gemini CLI
  • Cline
  • Amp
  • Roo Code
  • Goose
  • Continue
  • Kilo Code
  • Junie
  • Trae
  • OpenHands
  • Qwen Code
  • AI Agent
  • Tabnine Chat
  • Sourcegraph Cody
  • Amazon Q
  • Firebase Genkit
  • Sweep
  • Devin
  • Phind
  • Mentat
  • Claude Desktop
  • Aider
  • GPT Pilot
  • Plandex
  • bolt.new
  • v0
  • Replit Agent
  • Anse
  • TypingMind
  • BetterChatGPT
  • LibreChat
  • Warp AI
  • Pieces
  • Fig AI
Embed badge
vskill VERIFIED badge
[![vskill VERIFIED](https://verified-skill.com/api/v1/skills/karvifi/os/debug-systematic/badge)](https://verified-skill.com/skills/karvifi/os/debug-systematic)

Something wrong with this skill or its page? Tell us.