Evaluate proposed repository harness changes with pinned Darwin tools and explicit evidence limits.
The English summary is being prepared. Showing the original description.
See how to installReview the instructions the agent will receive before you install.
Run npm run bench:generate and npm run bench:verify from this package. The generated suite reflects repository tests; public, hidden and regression labels do not establish independent evaluation sets. Use a separately designed held out task set to support improvement claims.
npm run evolve uses the upstream real sandbox with repository tests. Upstream warns this substrate may score all mutated surfaces identically. npm run evolve:dry uses mock scores and is only a mechanics check. Neither proves improved agent capability. Configure the agent substrate and an explicit budget before model backed optimization. A successful gate is not permission to publish, deploy or merge beyond existing user authorization. Keep rollback evidence and reject regressions.
No reviews yet. Tell others what you thought.
No login needed. Share how it worked for you, ask a question, or note where it did not run.
No comments yet. Be the first to write one.
Inspect repository session replay and verify configured MetaHarness field memory boundaries before reusing outcomes.
The English summary is being prepared. Showing the original description.
Plan changes to this repository using its domain CLI, MCP boundaries, and harness validation evidence.
The English summary is being prepared. Showing the original description.