Keep improving one mission with evaluated trials.
Improves a single mission through bounded experiment cycles, preserving evaluator results and decision logs across iterations.
When to use it
Use when a mission and evaluator already exist from deep-interview and persistent improvement with durable experiment logs is wanted.
Give it an existing mission and evaluator in the project; it iterates on changes and saves evaluation results and decision logs until an explicit stop condition.
What you provide
Requires a mission and evaluator prepared beforehand through deep-interview --autoresearch; this skill does not generate the evaluator at runtime.
No additional actions listed.