Ephemeral GitHub Actions runners. Single binary, every platform. Containers on Linux, Hyper-V on Windows, VMs on macOS. Secure by default — every job gets an isolated, disposable environment.
AI-powered failure triage reporter for Playwright — classifies test failures (real bug / flaky / selector drift / env issue) and posts the verdicts to your PR
CI for Claude Skills — lint, eval, and regression-test SKILL.md files across a model matrix, with a self-growing eval loop that turns uncovered regressions into permanent test cases.
Drop it into your repo and let an AI Senior reviewer go through every PR — inline comments, severity-coded findings, and a general summary, all configurable per project and tech stack. Supports OpenAI, Anthropic, Gemini, and Ollama. Auto-detects your stack (NestJS, Next.js, React, TypeScript, Node, Flutter, Laravel) and applies built-in rules for