academic-research-agent

When to use

Use this skill when the user asks for academic research, literature review, systematic search, paper comparison, or synthesis of findings.

Core capabilities

1. Systematic literature search

   • Build keywords and boolean strings
   • Search scholarly sources (Scholar, IEEE Xplore, ScienceDirect, PubMed, ACM)
   • Define inclusion/exclusion criteria
   • Backward and forward citation chasing
   • Log search strategy for reproducibility

2. Source evaluation

   • Check relevance to research question
   • Assess methodological quality
   • Verify venue and author credibility
   • Identify bias and limitations
   • Prioritize by impact and recency

3. Structured data extraction (per paper)

   • Summary
   • Methodology
   • Results
   • Key findings
   • Limitations
   • Important concepts
   • Contributions
   • Implications
   • Further readings

4. Synthesis and analysis

   • Find patterns and trends
   • Identify gaps and contradictions
   • Cluster by themes
   • Compare methods and outcomes
   • Propose hypotheses or future directions

5. Documentation and reporting

   • Produce structured literature review
   • Build comparison tables
   • Format citations (APA/IEEE/Harvard)
   • Create annotated bibliography
   • Provide simple visuals or PRISMA flow

Workflow

Stage 1: Clarify research needs

Input: topic, research questions, goals
Output: research protocol (keywords, scope, criteria)

Stage 2: Systematic search

Input: research protocol
Output: candidate list with metadata

Stage 3: Screening and selection

Input: candidates
Output: final corpus with selection notes

Stage 4: Data extraction

Input: final corpus
Output: structured extraction table (CSV/Excel)

Stage 5: Synthesis and writing

Input: extraction table
Output: literature review draft with citations

Quality criteria

• Relevance: direct alignment with research questions
• Credibility: peer-reviewed preferred
• Recency: last 5 years unless seminal
• Traceability: clear link from source to claim

Output templates

Search report (markdown)

## Search Strategy
Database: Google Scholar
Query: ("cnn-lstm" OR "autoencoder") AND ("intrusion detection" OR "nids")
Date: 2026-02-02
Results: 847 hits
After filter: 234 hits

Extraction table (columns)

Author | Year | Methodology | Sample | Key Findings | Limitations

Synthesis structure (markdown)

## Theme 1: [Name]
[Synthesis with citations]

## Theme 2: [Name]
[Synthesis with citations]

## Research Gaps
[Specific gaps]

Limitations

• Cannot access paywalled sources without credentials
• Methodology appraisal still needs human validation

Usage example

research_topic: "Zero-day detection in NIDS using autoencoders"
research_questions:
  - "How effective are hybrid CNN-LSTM autoencoders for zero-day detection?"
  - "What datasets and metrics are used for cross-dataset validation?"
inclusion_criteria:
  - "Empirical studies with quantitative results"
  - "2019-2026"
exclusion_criteria:
  - "Non-technical reviews"
databases:
  - "Google Scholar"
  - "IEEE Xplore"
max_papers: 30
citation_style: "APA 7th"