Autoresearch: Autonomous Optimization Loop

Constraint + mechanical metric + fast verification = autonomous improvement.

Run N iterations that each make one atomic change, measure a single number, and keep or discard based on the score. Git is the memory and the rollback.

When to use

• Improve a measurable metric: test coverage, bundle size, ESLint error count, Lighthouse score.
• Autonomous execution over many iterations without manual steering.
• Git-tracked experiments where you want automatic rollback on regression.

When NOT to use