logic-checker

name	logic-checker
description	Detects logical discrepancies and inconsistencies in academic paper drafts, including claim-evidence mismatches, contradictions, and argumentation flaws.

Logic Checker Skill

This skill detects logical discrepancies, inconsistencies, and argumentation flaws in academic paper drafts following the CoPaper structure. It helps identify missing logical support, contradictions, and fallacies that weaken the paper's argumentation.

When to Use This Skill

• User requests to check logic in paper.md (e.g., "check logic in paper.md", "detect logical discrepancies")
• User wants to verify consistency between claims and evidence
• User wants to find contradictions within the paper
• User needs to validate argumentation quality

Input Files

File	Required	When to Read	Purpose
paper.md	Required	Step 1 (start)	Primary analysis target
.agents/state.json	Tool gap	Final step	Checker-result persistence is not currently exposed by the OpenCode plugin; do not edit state directly in plugin-based workflows

Do NOT read writingrules.md — the essential structure rules are inlined in the Paper Structure Reference section below.

Role and Responsibilities

You are an AI assistant performing logical analysis of academic papers. Your comments are AI-generated and must be clearly marked as such. Your analysis should be:

• Systematic: Check all argument structures throughout the paper
• Specific: Point to exact locations and provide concrete examples
• Constructive: Explain why the logic is problematic and how to fix it
• Evidence-based: Reference specific claims and supporting text
• Transparent: All comments must be explicitly marked as AI-generated to distinguish from human reviewer feedback

Key Markers and Their Meanings

Marker	Meaning	When to Use
<!-- AI Comments:	Start of AI-generated comment	ALWAYS use to begin every comment
**AI-GENERATED LOGIC ANALYSIS - FOR AUTHOR REVIEW**	Warning that content is AI-generated	ALWAYS include at the start of comment body
[DISCREPANCY TYPE]	Category of logical issue	ALWAYS include to classify the problem
[LOCATION]	Where the issue is found	ALWAYS include with section name and exact quote
[SEVERITY]	How serious the issue is	ALWAYS include (Critical/Major/Minor)
**END AI-GENERATED LOGIC ANALYSIS**	End of AI analysis content	ALWAYS include before closing -->

Logic Discrepancy Types

1. Claim-Evidence Mismatch

Definition: The evidence provided does not support the claim being made.

Type	Description	Example
Missing Evidence	Claim lacks supporting data or citation	"Method achieves high efficiency" without quantitative results
Mismatched Scope	Evidence is narrower than claim	Local results generalized to global conclusions
Insufficient Evidence	Evidence is too weak for claim strength	"Revolutionary improvement" based on marginal gains
Contradictory Evidence	Evidence actually contradicts claim	Claims novelty but cites identical prior work

2. Internal Contradictions

Definition: Two or more statements in the paper contradict each other.

Type	Description	Example
Direct Contradiction	Statement A says X, Statement B says not-X	"Results are significant" vs "Results were not significant"
Methodology-Result Mismatch	Method described doesn't match results reported	"10-fold cross-validation" in method, "5-fold" in results
Definition Inconsistency	Term used with different meanings	Variable defined one way, used differently later
Temporal Contradiction	Timeline inconsistencies	Claim about 2023 data, but citation is from 2021

3. Logical Fallacies

Definition: Flawed reasoning patterns that invalidate the argument.

Fallacy	Pattern	Detection Heuristic
False Cause	Correlation assumed as causation	"A increased, B increased, therefore A caused B"
Hasty Generalization	Insufficient sample for broad claim	Small dataset used for universal statements
Circular Reasoning	Conclusion assumed in premise	Claim restated as evidence for itself
False Dilemma	Only two options when more exist	"Either X or Y" when C, D are possible
Appeal to Authority	Expert cited outside domain	Non-expert cited for specialized claim
Straw Man	Misrepresenting opposing view	Oversimplified counterargument

4. Missing Logical Support

Definition: Gaps in the logical chain from claim to conclusion.

Type	Description	Example
Missing Warrant	No explanation of how evidence supports claim	Data presented without connecting logic
Unstated Assumption	Critical assumption not acknowledged	Claim depends on unstated prerequisite
Incomplete Reasoning	Logical steps skipped	Conclusion jumps without intermediate reasoning

5. Abstract-Body Consistency

Definition: Discrepancies between abstract claims and body content.

Type	Description	Example
Overstated Abstract	Abstract claims exceed body evidence	Abstract: "novel framework" → Body: incremental modification
Missing Abstract Claims	Important findings not in abstract	Key results mentioned only in body
Abstract-Method Mismatch	Abstract describes different work than method	Abstract: "deep learning approach" → Method: "statistical model"

Comment Structure

IMPORTANT: All comments generated by this skill are AI-generated analysis and suggestions. They must be clearly marked with "AI Comments:" to distinguish them from human reviewer feedback.

All Logic Check Comments must follow this standardized format:

<!-- AI Comments: 
**AI-GENERATED LOGIC ANALYSIS - FOR AUTHOR REVIEW**

[DISCREPANCY TYPE]
<Type from the 5 categories above>

[LOCATION]
Section: <section name>
Claim: "<exact quote of the problematic statement>"
<if applicable> Supporting Evidence: "<quote of evidence>"

[PROBLEM DESCRIPTION]
<explanation of why this is a logical problem>

[DETECTED ISSUE]
<specific description of the discrepancy found>

[SUGGESTED FIX]
<concrete suggestion on how to resolve the issue>

[SEVERITY]
Critical / Major / Minor
- Critical: Fundamental logical flaw invalidates argument
- Major: Significant gap that weakens argument
- Minor: Minor inconsistency that should be corrected

**END AI-GENERATED LOGIC ANALYSIS**
-->

Workflow

Step 1: Read Paper Structure

• Read paper.md to understand the paper's structure
• Identify the main sections and their purposes
• Note the hierarchical organization (Level 2-5 headers)

Step 2: Extract Argument Units

For each section:

1. Identify Level 6 headers (topic sentences/claims)
2. Extract supporting sentences (evidence)
3. Identify citations and references
4. Note any qualifiers, assumptions, or limitations mentioned

Step 3: Cross-Check for Consistency

Within-Section Checks:

• Compare each claim with its supporting evidence
• Verify warrants (logical connections) are stated or clear
• Check for internal contradictions

Cross-Section Checks:

• Compare abstract with body content
• Check methodology vs results consistency
• Verify introduction claims vs conclusion findings
• Cross-reference problem statement with evaluation

Step 4: Detect Logical Fallacies

For each argument:

1. Check causal claims for causation evidence
2. Verify generalization scope matches evidence scope
3. Detect circular reasoning patterns
4. Identify authority citations and verify domain relevance

Step 5: Generate Comments

For each discrepancy found:

1. Classify the type (from 5 categories)
2. Quote the exact location
3. Explain the logical problem
4. Provide specific fix suggestions
5. Assign severity level

Step 6: Insert Comments

• Place comments as HTML comments <!-- ... --> in the appropriate location
• Position them right before or at the relevant paragraph/section
• Ensure comments don't break document structure

Paper Structure Reference

The paper follows CoPaper structure. Key rules for this checker:

• Level 1-5 (# to #####): Structural headings only — no body text allowed under these.
• Level 6 (######): Content paragraphs. Title = topic sentence (≤50 chars). Body = supporting text (≤500 chars).
• Metadata: HTML comments <!-- description: ... --> guide what each section should contain.

Key Sections to Check

1. Insight Section: Is the insight clearly stated? Is there evidence supporting why this insight is valid? Are conditions for insight validity stated?
2. Problem Section: Is the problem clearly defined? Is the importance justified with evidence? Do scenarios match the problem description?
3. Existing Methods Section: Are limitations accurately described? Is there evidence that these are real limitations? Are root causes correctly identified?
4. Method Section: Does method address stated challenges? Are design decisions justified? Is there logical connection to the insight?
5. Evaluation Section: Do experiments validate the claims? Are baselines appropriate? Do metrics match the problem?

Output Format

Summary Report

After analyzing the paper, provide a summary:

## Logic Check Summary

**Paper**: [Paper title]
**Total Discrepancies Found**: X
- Critical: Y
- Major: Z
- Minor: W

### By Category:
1. **Claim-Evidence Mismatch**: X instances
2. **Internal Contradictions**: X instances
3. **Logical Fallacies**: X instances
4. **Missing Logical Support**: X instances
5. **Abstract-Body Consistency**: X instances

### Priority Fixes:
1. [Highest severity issue]
2. [Second highest]
3. [Third highest]

### Strengths:
- [What logical structures are well-done]

Inline Comments

Insert detailed HTML comments at problematic locations following the comment structure defined above. All comments must:

• Start with <!-- AI Comments:
• Include the marker **AI-GENERATED LOGIC ANALYSIS - FOR AUTHOR REVIEW**
• End with **END AI-GENERATED LOGIC ANALYSIS** before the closing -->

Examples

For detailed comment examples, see examples.md in this skill directory.

Common Patterns

Pattern 1: Checking Insight Correctness

For the Insight正确性分析 section:

• Check that conditions for insight validity are explicitly stated
• Verify each condition has supporting evidence (citation, theory, or data)
• Ensure "conditions not holding" scenarios are addressed
• Check that evidence for insight is not circular (assuming the insight to prove the insight)

Pattern 2: Checking Challenge-Solution Alignment

For the Insight挑战性分析 section:

• Verify each challenge is actually solved by the proposed method
• Check that "existing techniques cannot solve this" claims are justified
• Ensure solution approaches are logically connected to challenges

Pattern 3: Checking Evaluation Validity

For the 实验 section:

• Verify that experiments actually test the stated claims
• Check that baselines are fair comparisons (same task, same data)
• Ensure metric improvements are attributed to correct causes
• Verify "why baselines perform worse" explanations are logical

Pattern 4: Checking Abstract-Body Consistency

• Compare abstract claims with actual findings in the body
• Verify "novel", "first", "significant" claims are supported
• Check that methodology in abstract matches methodology section
• Ensure results in abstract match results in evaluation section

Important Notes

1. All comments are AI-generated: Every comment inserted by this skill is generated by AI analysis and must be clearly marked with "AI Comments:". These are NOT human reviewer feedback and should not be treated as such.
2. Be thorough but fair: Not every weak argument is a logical fallacy; distinguish between "could be stronger" and "is logically flawed"
3. Consider context: Some disciplines have different standards for evidence and argumentation
4. Preserve author voice: Comment on logic, not on writing style
5. Provide actionable feedback: Always suggest how to fix the issue
6. Prioritize by severity: Focus on critical and major issues first
7. Check cross-references: Many logical issues appear when comparing different sections
8. Verify citations: When claims depend on citations, verify the citation actually supports the claim
9. Authors should review all AI suggestions: AI-generated comments are suggestions for consideration, not definitive judgments. Authors should evaluate each suggestion critically.

Detection Checklist

Use this checklist during analysis:

Argument Structure

• Each Level 6 header (topic sentence) makes a clear claim
• Supporting sentences provide relevant evidence
• Logical connection (warrant) between claim and evidence is clear
• Scope qualifiers are appropriate

Consistency

• No contradictions between sections
• Abstract accurately represents body content
• Methodology matches results description
• Definitions are used consistently

Evidence Quality

• Quantitative claims have citations
• Citations actually support the claims made
• Sample sizes are appropriate for generalizations
• Causal claims have causation evidence

Fallacy Check

• No correlation-causation confusion
• No hasty generalizations
• No circular reasoning
• No false dilemmas
• Authority citations are domain-relevant