Research experiments that test your central claim.
Designs claim-driven research experiments and turns real results into comparison tables and evidence specifications.
When to use it
Use for experiment design, ablations, benchmark planning, result analysis, and evidence design within requested autonomous research.
Give it an experiment-design or result-presentation request; it returns the requested plan, table, or evidence-bound figure specification.
Must be activated first for every experiment task; its preflight is required even when no prose is produced.
Must be activated after ccf-humanization for every experiment task, with applicable rules reused or refreshed at handoffs.
Receives protocols, execution specifications, and interpretations during explicitly requested autonomous research and coordinates implementation and host execution; it is not required for ordinary experiment planning.
Resolves missing dataset, baseline, metric, or protocol provenance when a comparison depends on it; independent planning may continue without resolving that dependency.
No additional actions listed.