regression-test-from-bug

Converts bug reports into failing regression tests before writing the fix.

1|Updated Jul 3, 2026
One-click install
npx skills add https://github.com/Nandansai08/skillz --skill regression-test-from-bug-nandansai08
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: regression-test-from-bug
Source: https://github.com/Nandansai08/skillz/tree/main/skills/testing-qa/regression-test-from-bug
Command: npx skills add https://github.com/Nandansai08/skillz --skill regression-test-from-bug-nandansai08

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve? Bug fixes shipped without a failing-first test can silently regress, and post-fix tests written against already-fixed code prove nothing. This Skill enforces a disciplined workflow that turns every bug report into a permanent, minimal-layer regression test before the fix is written. ## Core Features & Use Cases - Reproduce-then-test workflow: Reduce the report to a minimal trigger, write the test at the lowest layer that exhibits the bug, and watch it fail for the right reason before fixing. - Sibling sweep and escape analysis: Grep for the same defect pattern elsewhere in the codebase and record which missing test category let the bug escape. - Verification checklist: Requires captured failing output, a green post-fix suite, ticket-linked test names, and documented sibling-sweep results. - Use Case: A customer reports being charged twice on retry. You minimize the repro to an idempotency-key bug, write a unit test at the payment client layer that fails showing two charges, fix the key derivation, and grep for the same pattern in refunds. ## Quick Start Use the regression-test-from-bug skill to turn this bug report into a failing regression test before I write the fix.

Frequently Asked Questions about regression-test-from-bug

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I write a regression test for a bug fix?▼

Reduce the bug report to a minimal failing input, write the test at the lowest layer that exhibits the bug, and confirm it fails with the reported wrong behavior before writing any fix. Name the test after the behavior and link the ticket in a comment.

Should I write the regression test before or after the fix?▼

Always before. A test written after the fix was never observed detecting the bug, so it proves nothing. The failing-first run is the evidence that the test actually guards the defect.

What layer should a regression test target?▼

The lowest layer that still exhibits the bug. If the broken logic is a pure function, unit test it there rather than writing a browser or end-to-end test just because the report came through the UI.

What if I cannot reproduce the bug?▼

No reproduction means no proof any fix works. Reproduction is the diagnosis; use bisection techniques to find the trigger when it is buried in a large input or commit range, and never close a ticket as fixed without a repro.

When should I not use a bug-driven regression test?▼

Do not use it for hypothetical edge cases found during normal test writing, which belongs to proactive edge-case enumeration, or for building first-time coverage on untested legacy code, which calls for characterization tests instead.