What problem does it solve? Research teams often brainstorm in ways that suffer from anchoring, groupthink, hidden assumptions, and untraceable decisions, producing candidate directions that cannot be reconstructed or defended later. This Skill provides a reproducible, evidence-aware workflow for generating, challenging, and prioritizing research directions while keeping ideas, assumptions, evidence, and decisions clearly separated. ## Core Features & Use Cases - Structured ideation workflow: Ten-stage process covering scoping, independent generation, clustering, criteria definition, adversarial review, literature checks, ethics gates, and decision logging. - Deterministic local CLIs: Standard-library Python scripts create session registers, validate structure and provenance, and compute a disclosed weighted scoring matrix with uncertainty intervals and weight sensitivity analysis. - Bias and integrity controls: Built-in guidance for production blocking, anchoring, authority effects, AI hallucination, dual-use screening, and responsible AI disclosure. - Use Case: A research group needs to prioritize candidate mechanisms for an observed phenomenon. They scaffold a session register, generate ideas independently, run adversarial review, check the literature, and produce a transparent weighted matrix whose output explicitly leaves the final decision to the accountable human owner. ## Quick Start Ask the agent to scaffold a scientific brainstorming session for your research question using the scientific-brainstorming skill, then generate ideas independently before any discussion or literature search.
