QA Forge started with a frustration from my day job: I was the only QA on a live product, and most of my week went into writing and re-running test cases that a machine should have drafted for me.

So I started building the tool I wished I had. QA Forge takes a feature description and generates candidate test cases, maps them against existing coverage, and suggests which regression paths actually need a re-run after a change. The goal is not to remove the human from QA. It's to stop the human from doing the boring 80% so the judgment calls get more attention.

MVP 1 is shipped. It does test case generation and a first pass at coverage analysis. The honest assessment: generation is useful immediately, coverage mapping is harder than I expected because real test suites are messy and don't map cleanly to features.

MVP 2 is where the CI/CD integration lives: running the regression suggestions inside the pipeline instead of as a separate step. That's what I'm working on now.

I'll keep posting build logs here as it progresses. If you're doing QA solo somewhere and this sounds useful, the repo is on GitHub. I'd genuinely like to hear what your worst repetitive testing task is.