autoresearch

You can automate autonomous iteration by defining a target metric, a verification command, and a commit policy. The automation proposes changes, validates outcomes, and keeps or rolls back commits to reach the measurable goal.

Constraint-driven autonomous iteration is a process that progresses toward a target metric by automatically proposing code changes and mechanically verifying results. It enables repeatable experiments for improving tests, benchmarks, and coverage.

You set up automated loops by providing an initial goal and a verifiable metric command. You can configure a user-defined iteration limit or run indefinitely while the tool logs results and handles rollbacks.

Yes, autonomous iteration can reduce build size by applying a policy for committing changes and handling rollbacks. It continuously proposes modifications, validates the build size metric, and decides whether to keep the outcome.

No specific testing framework is required, but you must provide a verifiable metric and a verification command. The tool uses mechanical verification to validate outcomes and log results for repeatable experiments.

Running indefinitely means the loop continues until a user-defined limit is reached or it is stopped manually. Without a limit, it relies entirely on the verification command and commit policy to manage continuous changes and rollbacks.