5 tagged falsification, measured the same way as everything else here.
Agent
Use this agent to turn one asserted limit, claim, or candidate root cause into a rigorously designed experiment — without running it. It returns a falsifiable hypothesis with an explicit null, a written probe artifact, a pre-committed outcome→conclusion table, and what each outcome unlocks. Spawn one per hypothesis…
Agent
Use this agent as the area chair who closes a peer-review round. It receives the submission packet, all independent reviews, the author rebuttal, and any re-scores, then issues the final decision (accept / minor-revision / major-revision / reject) with camera-ready requirements. It weighs evidence quality rather than…
Agent
Use this agent as one independent reviewer in a peer-review round for an invention, design, finding, or paper-style writeup. Each reviewer gets the same submission packet plus ONE assigned lens (soundness, prior-art/provenance, reproducibility, significance, or fatal-flaw) and works blind to the other reviewers.…