QA / test automation engineer. I care about tests that catch real regressions — not tests that just exist.
🚩 Flagship — assertkit
An AI-assisted QA toolkit with a human review gate on both halves:
assertkit generate— turn acceptance criteria into a Playwright spec, plus a review-notes file listing every assumption the model made. It never runs or commits the tests; I review and own what lands.assertkit triage— read a Playwright JSON report and classify each failure (real regression / flake / environment / assertion-too-strict) with a rationale and a next step. Runs in CI.
It also ships runnable patterns for testing non-deterministic LLM output without the tests turning flaky.
Playwright · Cypress · API & contract testing (node:test) · accessibility / WCAG audits · GitHub Actions CI · AI-assisted testing
Most of the smaller projects on my profile are scaffolded by craft-factory — a pipeline I built that ships a new QA project twice a week (Playwright suites, Cypress suites, API contract tests, accessibility audits), then reminds me to refine it. The scaffold is automated; the craft is done by hand.
Where LLMs genuinely help in QA: test generation from specs, failure triage, and how you assert on AI output without brittle, flaky checks.