idea-eval

Evaluates research ideas with a five-dimension rubric and feasibility score.

Updated Mar 27, 2026
One-click install
npx skills add https://github.com/megumi-ben/work17 --skill idea-eval-megumi-ben
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: idea-eval
Source: https://github.com/megumi-ben/work17/tree/main/.codex/skills/idea-eval
Command: npx skills add https://github.com/megumi-ben/work17 --skill idea-eval-megumi-ben

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve? Researchers often struggle to judge whether a concrete research idea is worth pursuing before investing in theory, experiments, and writing. This Skill applies a strict reviewer-style evaluation to decide whether an idea should be invested in, prototyped cheaply, reframed, or killed. ## Core Features & Use Cases - Five-Dimension Scoring: Rates novelty, theory-practice linkage, implementation cost, story quality, and hard-problem fit with weighted penalties. - Feasibility Score & Verdict: Produces a 0-100 Feasibility Score mapped to Invest, Cheap MVP First, Reframe, or Kill decisions. - Reviewer Simulation: Generates strongest accept/reject cases, top objections, and the most dangerous failing experiment. - Use Case: You have three candidate research ideas from a brainstorming session. Run this Skill to score each one, rank them, and identify the top candidate to pass into deeper development. ## Quick Start Evaluate this research idea strictly and give me a feasibility score with a final verdict: [paste your idea description].

Frequently Asked Questions about idea-eval

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I evaluate a research idea before starting experiments?

Submit your idea description to receive a structured evaluation covering novelty, theory depth, feasibility, story, and problem hardness. The output includes a 0-100 Feasibility Score and a verdict of Invest, Cheap MVP First, Reframe, or Kill.

What criteria should I use to judge research idea novelty?

Novelty is judged by the core contribution after removing surface packaging, not by terminology or framing. The evaluation asks what remains new, whether the community would care, and whether it solves a genuinely hard problem.

Can I evaluate multiple research ideas at once?

Yes, you can submit a list of candidate ideas. Each is scored individually, then ranked comparatively, with the top one to three survivors explicitly identified for further development.

When should I not use a research idea evaluation rubric?

Do not use it when your research direction is still vague or when you need idea generation rather than screening. It is designed for concrete, well-formed ideas, not open-ended exploration or rewriting weak ideas into strong ones.

Why does my research idea get a low feasibility score?

Low scores typically result from penalties like baseline treadmills, decorative theory, high ablation burden, or missing hard problems. Critical risks such as infeasible experiments or contributions that are merely known methods applied to new tasks also reduce the score.