Regression Testing

This handles the classic regression testing problem: you change one thing and need to know what might break without running your entire test suite for an hour. It analyzes git diffs to select affected tests, prioritizes by risk, and helps you set up tiered strategies like smoke tests on every commit and full suites nightly. The agents can generate regression tests from production bugs automatically, which is genuinely useful. The main insight here is smart selection catching 90% of regressions in 10% of the time. Watch out for transitive dependencies though, a model change three layers deep can still break a controller test that won't show up in naive diff analysis.

skills.sh installs: 334
GitHub stars: 524
last updated: Aug 21, 2026

View on GitHub

Install

$ npx -y skills add proffesor-for-testing/agentic-qe --skill regression-testing --agent claude-code

Installs into .claude/skills of the current project.

Files

Regression Testing

<default_to_action> When verifying changes don't break existing functionality:

ANALYZE what changed (git diff, impact analysis)
SELECT tests based on change + risk (not everything)
RUN in priority order (smoke → selective → full)
OPTIMIZE execution (parallel, sharding)
MONITOR suite health (flakiness, execution time)

Quick Regression Strategy:

Per-commit: Smoke + changed code tests (5-10 min)
Nightly: Extended regression (30-60 min)
Pre-release: Full regression (2-4 hours)

Critical Success Factors:

Smart selection catches 90% of regressions in 10% of time
Flaky tests waste more time than they save
Every production bug becomes a regression test </default_to_action>

Quick Reference Card

When to Use

After any code change
Before release
After dependency updates
After environment changes

Test Selection Strategies

Strategy	How	Reduction
Change-based	Git diff analysis	70-90%
Risk-based	Priority by impact	50-70%
Historical	Frequently failing	40-60%
Time-budget	Fixed time window	Variable