debug_skill
- Python
0
GitHub Stars
1
Bundled Files
2 months ago
Catalog Refreshed
4 months ago
First Indexed
Readme & install
Copy the install command, review bundled files from the catalogue, and read any extended description pulled from the listing source.
Installation
Preview and clipboard use veilstrat where the catalogue uses aiagentskills.
npx veilstrat add skill jjuidev/jss --skill debug- SKILL.md3.1 KB
Overview
This skill teaches a disciplined debugging workflow that enforces root cause analysis before any code changes. It combines systematic investigation, backward call-stack tracing, defense-in-depth validation, and strict verification so fixes are correct and durable. Use it to turn noisy symptom-chasing into reliable problem resolution.
How this skill works
The skill guides you through four phases: reproduce and gather evidence, analyze patterns against working examples, form and test minimal hypotheses, then implement a single, verified fix. When errors are deep in the stack it walks you backward to the originating polluter rather than patching symptoms. After fixing, it adds layered validations and requires fresh verification output before accepting completion.
When to use it
- Investigating test failures or flaky tests
- Diagnosing bugs or unexpected behavior
- Troubleshooting performance regressions or build/integration failures
- Tracing errors that appear deep in the call stack
- Before declaring work complete or merging a fix
Best practices
- Always reproduce the issue and collect concrete evidence before changing code
- Complete root cause investigation (Phase 1) before applying any fix
- Form minimal hypotheses and run targeted tests to validate them
- Fix at the source of the invalid data, not at the symptom layer
- Add validation at multiple layers (entry, business logic, environment, instrumentation) after fixing
- Run the exact verification commands and save their output before claiming success
Example use cases
- A failing CI test that passes locally — reproduce CI environment, trace pollution, and fix root cause
- A runtime exception surfaced in a downstream library — trace the call stack backward to the input that triggered it
- Intermittent performance degradation — collect traces, find pattern differences, test hypotheses with minimal repros
- A flaky integration test — bisect to find the polluting test and add defense-in-depth guards
- Finalizing a fix — run verification commands and attach fresh output to the change request
FAQ
Avoid blind quick fixes. If emergency patching is unavoidable, make the smallest targeted change, document evidence, plan a follow-up root cause investigation, and add verification once live.
How do I know I’ve found the root cause?
You have clear reproducible evidence, a minimal test that fails only with the suspected cause, and a successful targeted test that validates the hypothesis before implementing the permanent fix.
