inno-experiment-dev

Automating the ML experiment lifecycle involves coordinating planning, implementation, evaluation, and submission using integrated agents. This Skill orchestrates that complete workflow by applying automated feedback loops with checkpointing and reporting.

Structured experiment management applies coordinated planning and evaluation lifecycles to research projects. This Skill handles that by generating implementation plans, iterating software artifacts, and saving final model artifacts for reproducibility.

Reusing upstream repository-based resources during ML experiment planning is supported. The workflow coordinates Coding Plan, ML, and Judge Agents using reference codebases and prepared resources to generate detailed implementation plans.

Automated experiment management works without external dependencies in this setup. The metadata indicates an empty dependencies list, meaning the integrated judge and ML agent loop operates natively without requiring external package installations.

Applying judge agent feedback refines ML solutions through an automated iteration loop. This Skill builds and iterates software artifacts inside the active project workspace, directly applying the feedback to improve the model implementation.