iterative-experiment

What problem does it solve?

It eliminates the manual trial-and-error process of improving a measurable outcome by running a structured, repeatable loop that proposes changes, benchmarks them, and keeps only improvements.

Core Features & Use Cases

• Hypothesis-driven change loop: generates one scoped modification per iteration, commits it, measures performance, and keeps or reverts based on improvement.
• Baseline-first measurement: establishes a robust median baseline (with min/max variance context) before any experimentation begins.
• Reproducible experiment logging: persists a TSV log of every iteration (including commits and deltas) under a configured results path for later review.

Quick Start

Run iterative-experiment with a clear goal, a maximum budget, and an optional scope to let it iterate until your metric threshold is met or the budget runs out.