Launch evolutionary optimisation runs via scripts/run_evolution.py. Use when the user wants to evolve brain weights with CMA-ES or GA, evolve hyperparameters with CMA-ES or Optuna/TPE, or run evolution-framework smoke/pilot configs. Covers the timing pitfalls that crashed earlier sessions.
Run parallel experiment groups with multiple seeds. Use when the user wants to launch evaluation experiments across different configurations.
Analyse experiment results from simulation sessions. Use when the user wants to evaluate how experiments performed, diagnose issues, or decide next steps after running experiments.
Create or update an experiment logbook with artifacts, supporting data, and roadmap updates. Use when the user wants to document evaluation results permanently.
Comprehensive review of implementation before evaluation or merge. Checks completeness, correctness, code quality, test coverage, and more.
Review an OpenSpec change for completeness, correctness, and scientific rigor before implementation. Use before starting implementation of a change.
Update the evaluation scratchpad with latest experiment results. Use when the user wants to record findings from an evaluation round to the working scratchpad document.