setup

Setup is an agent skill that creates a new autoresearch experiment interactively. It collects everything an experiment needs: the domain, the target file, the evaluation command, the metric, the direction the metric should move, and the evaluator. You can run it with no arguments and answer the questions, or pass the values directly on the command line in order. It also lists existing experiments and the available evaluators, so you can check what is already set up before adding another. Getting the configuration right at the start matters, because the later iteration and loop skills rely on that metric and direction to judge each change. The skill is the first step in the autoresearch workflow of the claude-skills collection, followed by run for single iterations, loop for scheduled ones, and resume for paused experiments. Its security audit shows a warning from Gen Agent Trust Hub, and the other two scanners pass.…