Most Claude Code skills collections optimize for speed - get the feature built, ship the PR. Upstack takes the opposite approach: slow down, write the test first, then let Claude make it pass.
Released by Upsolve Labs, Upstack is a set of Claude Code skills built around red/green TDD (test-driven development, the practice of writing a failing test before writing any implementation code). The collection also includes skills for capturing screenshots and generating Postman collections - API testing files that let you verify your endpoints work correctly before deploying.
Where It Fits
The creators position Upstack as a complement to broader skills collections like gstack (from Y Combinator's Garry Tan). Their pitch: use the big-picture skills for the initial 80% of a project, then switch to Upstack for the polish phase where correctness matters more than velocity.
That's a distinction worth paying attention to. The current wave of AI coding tools has a well-documented weakness: they're great at generating plausible code but mediocre at generating correct code. A skills collection that forces Claude into a test-first workflow could help close that gap, at least for developers disciplined enough to use it.
The Practical Trade-Off
TDD with an AI assistant is an interesting tension. The whole point of using Claude Code is moving faster, but TDD deliberately adds friction to catch bugs early. Upstack is betting that the time you save by not debugging broken deploys outweighs the time you spend writing tests upfront. For production codebases, that's usually true. For quick prototypes, it's probably overkill.
The collection is open-source on GitHub, so you can cherry-pick the skills that fit your workflow rather than adopting the full TDD philosophy.