discussion-logic-diagnosis

该 Skill 针对英文心理学论文的 Discussion 进行逻辑诊断：检查前提—证据—结论链、因果与相关声明、数据/解释/推测三层区分、反方论点处理及结论范围，并输出结构化诊断报告；不生成或改写正文，参考文件引用尚待补齐。

laninga

热度 11.26

最近更新 2026-08-24 16:16:25

分类：数据依赖：纯提示词应用场景：研究 / 报告撰写

安装技能

复制下面的话术发给你的 Agent，它会通过 secure-skills 检索并安装该技能。

执行命令安装此 skill： npx secure-skills-cli add discussion-logic-diagnosis

Metadata

name: discussion-logic-diagnosis
description: Diagnose logical argumentation in the Discussion of an English psychology research paper. Checks argument chains (premise → evidence → conclusion), causal claims (does X cause Y, or correlate?), and the distinction between data, interpretation, and speculation. Triggers on "check logic", "argument check", "is my reasoning sound", "Discussion 逻辑", "讨论论证", "causal claim check", etc. Does NOT generate or rewrite prose.

Discussion Logic Diagnosis

Theory basis

Science Research Writing Unit 4.1.1 (forward-moving narrative wrap = logical thread) + 4.2.2 Q&A "What if I'm not confident about what my results mean?" (managing the data → interpretation → speculation chain) + 4.2.2 (mapping studies to literature is a logical act — placing your study in the research landscape).

Three-layer epistemic chain

Layer	What it is	Hedging required?
Data	What was directly observed	No (just describe)
Interpretation	What the data suggest	Some (the data "suggest", "indicate", "point to")
Speculation	What might be true beyond the data	Strong (it is "conceivable", "plausible", "possible")

A common error: collapsing layers — interpreting data with the same confidence as the data, or speculating as if interpreting.

What this skill diagnoses

Argument chain integrity — does each claim have visible support (premise → evidence → conclusion)?
Causal vs correlational claims — is "cause / lead to / result in" warranted by the design (experimental = OK; correlational = usually too strong)?
Data–interpretation–speculation layering — are these three levels kept distinct, with appropriate hedging at each?
Counter-argument handling — are limitations or alternative explanations acknowledged and addressed (not just listed)?
Generalisation scope — does the conclusion go beyond the sample/population the study actually used?
Internal consistency — does the conclusion follow from the premises, or does it introduce a new claim?

Diagnostic signals

Signal	Means
"Stress causes depression" (from a correlational study)	Causal overreach — "is associated with", "predicts"
"Therefore, .." (without a clear warrant from the prior sentence)	Missing premise — "therefore" without support
"Our results prove that..."	Data/interpretation collapse — studies rarely "prove"
Speculation introduced with the same hedging as data	Layer collapse — speculation should hedge more
Limitations listed in a final paragraph but never addressed in the argument	Surface-level concession — not real counter-argument handling
Conclusion in the last sentence introduces a claim not derived from the body	Internal inconsistency

Output format

## Logic Diagnosis

### Argument chain map
- [For each major claim: list premise, evidence, conclusion; flag any chain break]

### Causal claim audit
- [list all "cause / lead to / result in" with whether design supports]

### Epistemic layer audit
- [For each sentence: tag as data / interpretation / speculation; check hedging matches layer]

### Counter-argument handling
- [For each limitation: was it actually addressed in the argument?]

### Top issues (severity: critical / major / minor)
1. ...

Cross-reference

Causal claims also affect Conventions (causal overclaim is a conventions violation).
Take-home / forward motion is partly a logic concern but is owned by the Cohesion skill (Level 2 — global thread).

Academic integrity

Does not generate or rewrite. Only diagnoses.

References (to be filled in Phase 2)

references/rubric.md
references/checklist.md
references/examples/