Debug
Skill martonpaulo/skills/debug
My personal Agent Skills for coding agents: engineering workflows plus a few personal-use tools. Personalized forks, with every upstream author credited.
npx -y skills add martonpaulo/skills --skill debugAssembled from the repository path, not quoted from the project. Check it against their README if it does not work.
2 things to look at
- 14 days oldThe repository was created 14 days ago. New is not bad, but a brand new repository carrying a familiar-sounding name is the shape a typosquat arrives in, and there has been no time for anyone else to find a problem with it.
- 0 stars0 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.
What its author says it does
Copied from the file, not written here
Diagnose a non-trivial bug through reproduction, evidence gathering, hypothesis testing, root-cause analysis, a minimal fix, and regression verification. Use when the cause is not already obvious.
SKILL.md
2.2 KB, as published. Nobody here has run it
Debug
Use this discipline for difficult bugs, intermittent failures, and performance regressions. For a local failure with an already-proven cause, use the normal implementation workflow.
When a hypothesis depends on documented API behavior, version compatibility, deprecation, configuration semantics, or migration changes, use apple-docs for Apple development or deep-docs elsewhere. This skill retains ownership of reproduction, diagnosis, and the fix.
Workflow
- Inspect the relevant code, call sites, tests, logs, and recent changes before editing.
- Reproduce the reported symptom or gather the strongest available concrete evidence. Confirm that the evidence matches the user's bug rather than a nearby failure.
- Reduce the scenario while preserving the failure when practical. Record the smallest reliable reproduction command or procedure.
- Separate symptoms from possible causes. Write explicit, falsifiable hypotheses and rank them by evidence.
- Test one hypothesis at a time with targeted instrumentation, controlled inputs, a debugger, a profiler, or a focused test. Remove temporary instrumentation afterward.
- Identify the root cause and make the smallest correct fix. Avoid batches of speculative or unrelated changes.
- Add a regression test at the closest stable behavioral seam when practical.
- Run the focused reproduction or test first, then broader relevant anti-regression checks.
Evidence gaps
If reproduction is impossible, report what was tried, what evidence is missing, and what artifact or access would improve confidence. Do not present a hypothesis as a confirmed cause.
Completion
Claim the bug is fixed only when the original symptom no longer reproduces, focused verification passes, broader checks show no relevant regression, and temporary debugging changes are gone. Report the root cause, fix, commands run, results, and residual uncertainty.