zms618/research-idea-discovery

Mandatory final mentor-style, plain-language, experiment-ready Idea report after Research Idea Discovery screening, including scientific motivation, defensible contributions, story, linked datasets/models/baselines and E0/weekly plan, reviewer attacks and kill criteria.

Use when Research Idea Discovery cannot identify a defensible research idea after evidence-based exploration, when important related-work or dataset evidence is missing, or when a user supplies a Deep Research report for continuation. Produces a tailored copyable research prompt and verifies imported evidence without claiming to launch Deep Research directly.

Convert a selected research hypothesis into an accessible-dataset and model-based validation plan with compatible baselines, 48-hour E0 pilot, first-week steps, resource evidence, and explicit kill criteria. Use after evidence and feasibility investigation.

Use when user requests research hypothesis discovery candidate generation mechanistic gap analysis. Works within the general Research Idea Discovery plugin and any user-selected field.

Use for finding, validating, refining, comparing or rejecting scientific research ideas and paper contributions in any user-chosen research field. Handles literature-grounded gap discovery, mechanism-level novelty checks, falsifiable hypotheses, minimum decisive experiments, and reviewer-style critique. Target venue (including CCF-A) and research domain are selectable, never fixed. Do not trigger for ordinary paper summarization, writing polish, or generic brainstorming unless the user asks to discover or evaluate novel research contributions.

Translate a rigorously investigated research idea into intuitive motivation, a concrete scenario, a two-case example, and a causal story a graduate student can retell. Use as part of Research Idea Discovery final delivery or for requests like 我没懂这个idea.

Use when user requests paper literature retrieval evidence mapping query planning and source verification. Works within the general Research Idea Discovery plugin and any user-selected field.

Design and adversarially validate an implementable scientific method from a research motivation or selected idea. Requires operational estimators, interfaces, objectives, update timeline, strongest simple alternatives, code integration plan and decisive experiments; avoid module stacking and unsupported claims.

Generate and compare several independently grounded research motivations before researching innovation points. Use when targeting major vision conferences or CVPR best-paper aspirations, scouting multiple scientific problems, or selecting a few defensible directions for targeted gap research.

Use for feasibility-gated research motivation strength auditing, scientific storytelling, reviewer persuasion tests and evidence-backed narrative revision for any user-selected research field. It must not fabricate importance, experiments or novelty.

Use when user requests prior art novelty scooping collision auditing scientific research papers. Works within the general Research Idea Discovery plugin and any user-selected field.

Critically read a small adversarial group of anchor papers to discover real, falsifiable research problems from failures, successful-method mechanisms, contradictions and deployment assumptions; reuse verified assets without blindly modifying the previous network.

Use when user requests falsification pilot experiment design hypothesis testing research planning. Works within the general Research Idea Discovery plugin and any user-selected field.

Use at the beginning of scientific idea discovery to find real, important, falsifiable research failures from actual sources; reject weak motivations before architectures and novelty-hunting. Includes low-cost problem-evidence scouting and a hard Motivation-First Gate.

Use when user requests adversarial peer review scientific claim audit independent reviewer. Works within the general Research Idea Discovery plugin and any user-selected field.