autoresearch

Iterative improvement for a single mission is automated by cycling through stateful evaluation and iteration loops governed by strict evaluator contracts. The process evaluates, records outcomes, and repeats within max-runtime controls to ensure continuous enhancement.

Stateful single-mission iterative improvement is a continuous evaluation process that records mission progress and repeats enhancements based on known contracts. You need it when refining specific missions against strict requirements and durable logging.

You track mission improvements using structured JSON evaluator outputs paired with human-readable markdown logs. These formats provide comprehensive decision tracking and machine-readable logging for monitoring iterative enhancement cycles.

Yes, strict evaluation constraints are enforced during mission iteration to ensure improvements adhere to contract terms. The evaluator contracts validate each iteration cycle against known requirements before recording and repeating.

Max-runtime control for single-mission improvement loops is configured to limit the duration of iterative evaluation cycles. This boundary prevents runaway processes while maintaining stateful iteration and durable logging within defined limits.

Yes, autoresearch operates without external dependencies, relying solely on its internal scripts, references, and assets. This self-contained structure enables stateful evaluation loops and markdown logging without requiring additional packages.