yanacuti1121/yana-ai/ai-regression-testing

Trust tier 3
  • community
Source check pending

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

Write regression tests targeting AI-introduced bugs and blind spots. Use when "a bug was found and fixed — prevent re-introduction", "AI agent modified API routes or backend logic", "running /bug-check", "we keep seeing the same type of bug come back", "AI keeps making the same mistake". Do NOT use for general test coverage — this skill targets specific known failure modes only.

Install

npx vskill@latest install yanacuti1121/yana-ai/ai-regression-testing

SKILL.md

When to Use

  • Use when: an AI agent introduced a bug that was then manually fixed
  • Use when: the same type of bug appears repeatedly across sessions
  • Use when: code was refactored by an agent and existing behavior might have changed
  • Do NOT use: for covering new features before they're built (that's TDD, not regression)

Core Principle

Don't aim for 100% coverage. Aim for 100% coverage of known failure modes.

AI models have predictable blind spots. Write tests that specifically catch what AI gets wrong repeatedly.

Common AI Blind Spots to Cover First

1. Error state leakage — happy path tested, error path returns wrong data
2. Missing rollback — DB write succeeds, second step fails, no rollback
3. SELECT clause omissions — query returns partial data silently
4. Sandbox/production path divergence — works in test, breaks with real env vars
5. Auth check on create but not update/delete — classic AI shortcut
6. Off-by-one in pagination — AI copies the pattern, gets limits wrong
7. Unhandled null from optional fields — AI trusts the type, skips null guard

Regression Test Workflow

Bug found in /api/user/profile
↓
Write a test that FAILS with the bug present
↓
Verify the test actually fails (run it now)
↓
Apply the fix
↓
Verify test now passes
↓
Commit test + fix together

The test-fails-first step is non-negotiable. A test that passes before the fix doesn't prove anything.

Setup: Sandbox Mode

Force tests to run without real infrastructure:

typescript
// vitest.config.ts — force sandbox mode for all tests
export default defineConfig({
test: {
env: { NODE_ENV: 'test', SANDBOX_MODE: '1' },
setupFiles: ['./tests/setup.ts'],
},
})
// tests/setup.ts
process.env.DATABASE_URL = undefined // no real DB
process.env.SANDBOX_MODE = '1'

Example: Regression Test Pattern

typescript
// Bug: /api/user/profile returned 200 with empty object when user not found
describe('GET /api/user/profile regression', () => {
it('returns 404 when user does not exist', async () => {
const res = await request(app)
.get('/api/user/profile')
.set('Authorization', 'Bearer valid-but-unknown-user-token')
expect(res.status).toBe(404) // was 200 before fix
expect(res.body.id).toBeUndefined() // was {} before fix
})
})

Anti-Fake-Pass

❌ Writing the test after the fix (test may never have caught the bug)
❌ Test that passes before AND after the fix (proves nothing)
❌ Sandboxed test that never hits the actual auth/DB path the bug was in
❌ "We trust the AI won't make that mistake again" — it will
✅ Run test with bug present → confirm FAIL → apply fix → confirm PASS
✅ Tag regression tests: describe('[regression] bug description', ...)

Security

  • Security scanPassed, 90 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
0

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/yanacuti1121/yana-ai/ai-regression-testing/badge)](https://verified-skill.com/skills/yanacuti1121/yana-ai/ai-regression-testing)

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