skills

Why we included this project

This is a library of small, composable agent skills that turns a single coding agent into something closer to a small autonomous engineering team. The skills walk through the full lifecycle of a feature: mapping the unknowns up front, writing a spec that splits the work into independently verifiable slices, implementing in reviewable passes, then closing with code review, visual checks, and a record of the decisions the agent made on your behalf. Each skill stands alone, so you can chain the whole loop for a large feature or pull out just one when the AI touches code ad hoc. The part that earns its keep is the review discipline: the run ends by surfacing the choices the agent made where the spec was silent, so you audit decisions instead of reading a giant diff. That makes it a good fit for developers who want longer, more autonomous agent sessions without giving up human oversight.