plan

Create an agent team to critique this plan $ARGUMENTS.

This covers both early ideas and concrete designs — bring whatever you have.

Lead: present and defend the idea or design. Steel-man it. Skeptic: find all prior work that overlaps. Is this actually new? Are we testing what we think we're testing? Methodologist: evaluate feasibility, identification strategy, threats to validity, statistical power. If it's still an idea, flag what would need to be true for it to work. Visionary: is this ambitious enough? Is the framing big enough to matter? Newcomer: flag where the idea or design is unclear. If you can't follow the logic, that's a problem. Connector: are there related approaches or methodological innovations from adjacent fields?

Then argue. Produce an assessment: go/no-go if it's an idea, validated design or specific revisions if it's a methodology.