Description
Optimizing code means edit, benchmark, repeat, and you only test a few ideas a day. evo turns your codebase into an autoresearch loop, discovering what to measure, instrumenting benchmarks and running tree search with parallel subagents.
It works with Claude Code and Codex, starts in two commands and includes a dashboard for every experiment.
Auto benchmarks:Metrics discovered.
Tree search:Parallel subagents.
Platforms:Claude Code and Codex.
Dashboard:Experiments visualized.
It works with Claude Code and Codex, starts in two commands and includes a dashboard for every experiment.
Features
Auto benchmarks:Metrics discovered.
Tree search:Parallel subagents.
Platforms:Claude Code and Codex.
Dashboard:Experiments visualized.
