autoresearch

Automate prompt optimization by running an autonomous iteration loop that continuously refines prompts against measurable criteria. The process requires Python to generate eval.py or rubric.md, establishing a baseline before iteratively testing and modifying inputs to improve outcomes.

Yes, you can optimize code and configurations autonomously without manual intervention. The system supports an autonomous iteration loop that continuously refines templates and code against measurable criteria, utilizing deterministic and AI judge modes for evaluation.

Deterministic and AI judge modes offer different evaluation approaches for template refinement. Deterministic mode uses fixed criteria, while AI judge mode leverages an AI agent. Both ensure clear separation between the optimizer, eval agent, test runner, and judge for fairness.

Yes, you need Python to run autonomous optimization for prompts and code. The framework requires Python for generating eval.py or rubric.md files, and may require external libraries to execute specific testing and evaluation tasks during the iteration loop.

Track progress during autonomous code optimization by using the built-in dashboard, which provides a visual representation of the optimization progress and results. You can also optionally integrate Git for version control during the process.

Use an autonomous iteration loop instead of manually refining prompts to achieve continuous improvement without manual intervention. It enforces clear, measurable, and testable criteria while separating roles between optimizer and judge for security and fairness.