autoresearch

Bounded iteration is an autonomous optimization loop design that tracks iterations, baselines, and results with git-backed history. This ensures mechanical verification and transparent traceability for reproducible debugging and security audits.

To start an autonomous optimization loop, invoke the improvement command to trigger iterative prompts and validations. This initiates continuous refinement across multiple agents, applying mechanical verification and tracking results with git-backed history for reproducibility.

You should use an autonomous optimization loop when your task requires iterative optimization, such as software improvement, debugging, or security audits. It is ideal for complex task orchestration that benefits from repeatable loops and continuous refinement.

Autonomous optimization loops are limited by bounded iterations, meaning they enforce a strict maximum number of runs to prevent infinite loops. You should anticipate these constraints when planning complex task orchestration to ensure mechanical verification completes within set boundaries.