What it does
A plugin that runs optimization experiments on a codebase. It finds what is worth measuring, sets up the benchmark, then tries changes in a loop, keeping those that improve the score and dropping the rest. It uses tree search with parallel subagents and includes a dashboard to follow the runs.
Use cases
- 01Find and measure a performance metric in a project
- 02Run many code variants and keep the best scoring one
- 03Follow experiment results in the dashboard