drchronx/ai-agent-research-starter-kit/academic-paper-strategist

Academic Paper Strategist

Overview

This skill provides a systematic framework for strategic planning of academic papers in philosophy and interdisciplinary research. It guides you through three phases—from platform selection to optimized outline—with AI-driven literature search, research gap identification, originality assessment, and quality-controlled outline design.

Output: A detailed, review-ready paper outline with supporting documentation (platform style guide, literature review, gap analysis, reviewer assessment).

Companion Skill: Use academic-paper-composer to execute the outline and write the full paper.

When to Use This Skill

Use academic-paper-strategist when you need to:

Planning Stage:

• Design a research paper from initial idea to structured outline
• Identify a suitable publication platform (PhilArchive, arXiv, etc.)
• Understand writing standards for a specific preprint platform

Research Stage:

• Conduct systematic literature search
• Identify research gaps with evidence
• Assess originality of your research idea
• Predict potential impact

Design Stage:

• Structure paper chapters and arguments
• Optimize outline from reviewer perspective
• Prepare submission-ready strategy

Triggers:

• "Plan a paper on [topic]"
• "Help me design a paper about [subject]"
• "Identify research gaps in [field]"
• "Is this idea original?"
• "What platform should I submit to?"

Workflow Overview

Phase 1: PLATFORM ANALYSIS (Target Selection + Style Learning)
    ↓
Phase 2: THEORETICAL FRAMEWORK (AI-Driven Gap Identification)
    ↓
Phase 3: OUTLINE OPTIMIZATION (Quality-Controlled Design)
    ↓
Output: Detailed Outline + Supporting Documentation

Quality Gates: 3 validation checkpoints ensure each phase meets standards before proceeding.

Phase 1: Platform Analysis

Goal

Identify the optimal submission platform and understand its writing standards through systematic sample paper analysis.

Input Required from User

• Core research idea or topic (brief description)
• Target platform (optional - if unclear, I'll recommend)
• Field/discipline (philosophy, cognitive science, interdisciplinary, etc.)

Workflow

Step 1.1: Platform Selection (If Needed)

If target platform unclear, I will:

1. List candidate platforms based on research content:

• PhilArchive/PhilPapers: Philosophy papers, phenomenology, metaphysics
• arXiv (cs.AI, q-bio.NC): Computational, neuroscience, AI-related
• PhilSci-Archive: Philosophy of science, formal methods
• PsyArXiv: Psychology, cognitive science
• SocArXiv: Social sciences, interdisciplinary

1. Evaluate each platform:

• Subject area alignment (does your topic fit?)
• Methodology match (philosophical/empirical/computational)
• Acceptance criteria
• Typical review timeline

1. Provide recommendation with reasoning

1. Decision Point 1: You confirm platform or suggest alternative

Step 1.2: Sample Paper Search (AI-Driven, Quality-Controlled)

I will conduct multi-dimensional search for 8-10 representative papers:

Search Strategy (load references/search_strategy.md for details):

Time Dimension:

• Recent (last 6 months): 3 papers - capture current trends
• Current (1-2 years): 3 papers - established standards
• Classic (highly cited): 2 papers - quality benchmarks

Relevance Dimension:

• Use keyword combinations from your topic
• Score each paper 0-10 for relevance
• Retain only papers scoring ≥7/10

Diversity Dimension:

• Multiple authors (≥5 unique)
• Different research perspectives
• Varied paper lengths

Tools Used:

• Exa MCP (semantic search)
• Tavily MCP (web search)
• Platform-specific search (PhilPapers, arXiv)

Quality Validation: After search, I'll run scripts/evaluate_samples.py to generate evaluation report:

python scripts/evaluate_samples.py

This produces:

• Sample quality metrics
• Time distribution check
• Relevance statistics
• Diversity assessment
• Pass/Fail recommendation

Quality Gate 1 (Must Pass):

• ✓ Sample papers ≥8
• ✓ Time distribution balanced
• ✓ Average relevance ≥8/10
• ✓ Unique authors ≥5

If Failed: Re-search with adjusted criteria

Step 1.3: Writing Standards Extraction

From the 8-10 sample papers, I will extract:

Structural Patterns:

• Abstract structure (Problem→Method→Results→Contribution?)
• Chapter organization (how many sections? typical flow?)
• Average proportions (Intro 15%, Main 70%, Conclusion 15%?)

Style Patterns:

• First-person vs passive voice usage
• How arguments are structured
• Citation density and format
• Use of technical terminology

Format Specifications:

• Typical word count range
• Reference count range
• Section heading conventions

Output: [Platform]_Writing_Standards_Guide.md

Phase 2: Theoretical Framework

Goal

AI-driven systematic literature search, research gap identification, and originality assessment.

Input Required from User

• Core research question/thesis (your main argument)
• Background context (why you're interested in this)
• Optional: Any papers you already know about

Workflow

Step 2.1: Literature Search (AI-Driven, Fully Automated)

Important: This phase is AI-driven. You provide your idea; I conduct comprehensive literature search and gap analysis.

Multi-Round Search Strategy:

Round 1: Direct Search (Primary Literature)

1. Extract core concepts from your idea (3-5 concepts)
2. Generate keyword combinations (10-15 combinations)

• Concept + concept
• Concept + method
• Include synonyms and disciplinary variants

1. Search each combination using Exa/Tavily
2. Collect 30-50 candidate papers
3. Quality filter: Retain top 20 papers (relevance ≥7/10)

Round 2: Expanded Search (Adjacent Fields)

1. Extract new keywords from Round 1 papers
2. Search adjacent disciplines:

• Philosophy → cognitive science
• Neuroscience → philosophy of mind
• AI → consciousness studies

1. Collect 10-20 bridging papers

Round 3: Classic Literature (Foundational Works)

1. Identify highly-cited papers (>100 citations)
2. Track citations from Round 1-2 papers
3. Collect 5-10 foundational papers

Total Literature Base: 35-50 papers

Load Reference: references/search_strategy.md for detailed methodology

Step 2.2: Research Gap Identification (AI Analysis)

Using collected literature, I will automatically identify 3-5 research gaps:

Gap Identification Methods:

1. Concept Mapping:

• Plot papers on Concept × Method matrix
• Identify white spaces (unexplored combinations)

1. Problem-Solution Analysis:

• What problems does literature address?
• What limitations do authors acknowledge?
• What questions remain unanswered?

1. Temporal Analysis:

• What was once studied but abandoned?
• What emerged recently but unexplored?

Gap Types:

• Complete gaps: No existing research
• Partial gaps: Preliminary work only, needs development
• Controversy gaps: Competing theories, no resolution

For Each Gap, I Document:

• Clear definition (50-100 words)
• Evidence (3-5 citations showing gap exists)
• Significance assessment (High/Medium/Low)
• Feasibility assessment (Can you address it?)

Validation: Run scripts/gap_analysis.py to ensure quality:

python scripts/gap_analysis.py

This validates:

• Each gap has ≥3 pieces of evidence
• Definitions are specific and clear
• Significance is justified

Quality Gate 2 (Must Pass):

• ✓ Literature base ≥20 papers
• ✓ Identified gaps ≥3
• ✓ Each gap has ≥3 evidence citations
• ✓ At least 1 high-significance gap

If Failed: Continue search or pivot research direction

Output: Literature_Review_Report.md + Research_Gap_Analysis.md

Step 2.3: Originality Assessment (AI Analysis)

I will automatically assess your idea's originality:

Step 1: Similarity Analysis

• Compare your idea with top 15 most similar papers
• Create similarity matrix (topic/method/conclusion overlap)
• Calculate overall similarity percentage

Interpretation:

• >80%: High similarity, needs repositioning
• 50-80%: Moderate, emphasize differences
• <50%: Good originality, proceed

Step 2: Innovation Classification

Identify which innovation types apply (need ≥2):

1. Methodological: New approach to known problem
2. Theoretical: New framework or model
3. Application: Existing theory to new domain
4. Integrative: Synthesizing separate literatures

Step 3: Impact Prediction (1-10 scale)

Scoring Criteria:

• Gap Importance (5 points): Core vs. peripheral problem?
• Generalizability (3 points): Widely applicable?
• Explanatory Power (2 points): Resolves existing puzzles?

Target: ≥7/10 for good impact potential

Output: Originality_Assessment_Report.md (similarity analysis + innovation types + impact prediction + 300-word justification)

Step 2.4: Core Concepts Discussion (Interactive)

Decision Point 2: Based on literature analysis, I will:

1. Propose 3-5 core concepts to emphasize
2. Explain rationale (based on gap analysis + literature frequency)
3. Ask for your feedback: Agree? Adjust? Add?

This ensures the paper focuses on the right concepts to maximize contribution.

Phase 3: Outline Optimization

Goal

Design a structured, review-ready outline optimized from a reviewer's perspective.

Input

• Literature analysis from Phase 2
• Core concepts (confirmed in Step 2.4)
• Platform standards from Phase 1

Workflow

Step 3.1: Initial Structure Design

Based on platform standards, I will:

1. Design chapter structure:

• Abstract
• Introduction (with subsections)
• Main body (3-5 chapters, each with subsections)
• Conclusion

1. Allocate word counts:

• Introduction: 15-20% of total
• Main body: 60-70% of total
• Conclusion: 10-15% of total

1. Determine argument flow:

• Logical progression of ideas
• Where to introduce concepts
• Where to address objections

Output: Initial_Outline_Draft.md

Step 3.2: Reviewer-Perspective Self-Assessment

I will evaluate the outline as if I were a platform reviewer, using 7 dimensions (load references/quality_standards.md for criteria):

7-Dimension Assessment (5 points each, 35 total):

1. Argument Clarity (1-5)

• Is the thesis clear?
• Are supporting arguments identifiable?

1. Argument Completeness (1-5)

• Any logical gaps or jumps?
• All premises justified?

1. Literature Support (1-5)

• Expected citation count (40+ for philosophy)
• Key works covered?

1. Methodological Clarity (1-5)

• Approach explicit (philosophical argument/phenomenological/etc.)?
• Method justified?

1. Originality Expression (1-5)

• Contribution clear?
• Differentiated from existing work?

1. Organization (1-5)

• Logical flow?
• Proportions balanced?

1. Platform Fit (1-5)

• Matches platform style?
• Meets format requirements?

Scoring:

• Total: X/35
• Passing threshold: ≥28/35 (80%)

Requirement: Must identify at least 3-5 specific issues with concrete improvement suggestions.

Output: Reviewer_Assessment_Report.md

Step 3.3: Optimization Recommendations (Data-Driven)

For each dimension scoring <4/5, I will provide:

Issue Description:

• What specific problem exists?

Severity (High/Medium/Low):

• High: Affects paper acceptability
• Medium: Affects paper quality
• Low: Minor improvement

Concrete Solution:

• Specific actionable fix
• Example of how to implement

Expected Improvement:

• How much will this raise the score?

Prioritization:

1. All high-severity issues first
2. Then medium-severity
3. Then low-severity (optional)