Agent Skills
› KunAgent/Kun
› scientific-problem-selection
scientific-problem-selection
GitHub评估和优先排序研究问题，基于重要性、新颖性、可行性、证据、资源、伦理和可证伪性。通过工具路由检索文献并评分，辅助科研选题决策。
触发场景
需要评估研究问题的优先级
进行科研选题决策
比较候选研究问题
安装
npx skills add KunAgent/Kun --skill scientific-problem-selection -g -y
SKILL.md
Frontmatter
{
"name": "scientific-problem-selection",
"description": "Evaluate and prioritize research questions by importance, novelty, tractability, evidence, resources, ethics, and falsifiability."
}
Scientific Problem Selection
Copyright (c) 2026 KunAgent. Licensed under the MIT License.
Purpose
Evaluate and prioritize research questions by importance, novelty, tractability, evidence, resources, ethics, and falsifiability.
Tool routing
| Tool or skill | Use |
|---|---|
web_fetch |
Retrieve public papers, datasets, calls, or standards. |
browser_use |
Inspect public literature and research portals. |
render_chart |
Compare candidate problems using structured criteria. |
load_skill |
Load the scholarly-research skill for a dedicated literature review. |
Workflow
- Define field, decision horizon, available methods, resources, and constraints.
- Express each candidate as a falsifiable question with a measurable outcome.
- Map current evidence, competing explanations, and decisive experiments.
- Score importance, novelty, tractability, risk, ethics, and learning value.
- Choose a portfolio with a lead problem and fallback path.
Completion gates
- Verify novelty claims against current literature.
- Identify required sample size, data access, and ethical approvals.
- State what result would change the research direction.
Boundaries
- Do not equate fashionable topics with important questions.
- Do not recommend work that lacks a plausible ethical and resource path.
Delivery
Lead with the outcome, name the evidence used for verification, and disclose any real limitation that remains.
版本历史
- a984912 当前 2026-09-09 01:50
