perplexity-academic

name	perplexity-academic
description	Academic-focused web search optimized for scholarly research. Pre-configured with academic mode, peer-reviewed source prioritization, and curated academic domains. Use for literature searches, citation verification, and research discovery.

Perplexity Academic Search

Overview

A specialized academic search skill built on Perplexity's Academic Mode. This skill comes pre-configured with optimal settings for scholarly research:

• Academic mode enabled by default - Prioritizes peer-reviewed sources
• Curated academic domains - Nature, Science, Cell, NEJM, PubMed, arXiv, etc.
• High context size - Maximum detail for comprehensive literature searches
• Date filtering - Focus on recent research (configurable)

For general web search, use perplexity-search instead. For full Perplexity features, see .claude/docs/perplexity-best-practices.md.

When to Use This Skill

Use this skill when:

• Searching for peer-reviewed academic papers
• Finding recent scientific publications
• Conducting literature reviews
• Verifying claims against scholarly sources
• Discovering research in specific academic domains
• Supporting systematic reviews

Do not use for:

• General web searches (use perplexity-search)
• Technical documentation (use --preset technical)
• News and current events (use --preset news)
• Non-academic fact-checking

Quick Start

Setup

Uses the same API keys as perplexity-search:

# In .env file - Direct API (preferred)
PERPLEXITY_API_KEY=pplx-your-key-here

# OR OpenRouter fallback
OPENROUTER_API_KEY=sk-or-v1-your-key-here

Note: Academic mode requires Direct Perplexity API. OpenRouter has limited feature support.

Basic Usage

# Simple academic search (academic mode auto-enabled)
python scripts/perplexity_academic.py "CRISPR gene editing clinical trials"

# With date filter for recent research
python scripts/perplexity_academic.py "transformer architecture improvements" --recent 2023

# Medical/clinical research
python scripts/perplexity_academic.py "mRNA vaccine mechanisms" --domain medical

# With Pro Search for complex analysis
python scripts/perplexity_academic.py "Compare BERT vs GPT architectures" --pro

# Save results with citations
python scripts/perplexity_academic.py "gut microbiome Parkinson's" --output results.json

Pre-Configured Settings

This skill automatically applies:

Setting	Value	Purpose
search_mode	academic	Prioritize peer-reviewed sources
context_size	high	Maximum detail for research
model	sonar-pro	Best balance of quality/cost
max_tokens	4000	Comprehensive responses

Default Academic Domains

When no --domain specified, searches across (16 domains, under API limit of 20):

arxiv.org, nature.com, science.org, cell.com,
nih.gov, pubmed.gov, ncbi.nlm.nih.gov, pnas.org,
sciencedirect.com, springer.com, wiley.com, plos.org,
biorxiv.org, medrxiv.org, .edu, .gov

CLI Options

Usage: perplexity_academic.py <query> [options]

Positional:
  query                    Academic search query

Domain Presets:
  --domain DOMAIN         Domain preset: general, medical, biomedical,
                          cs, physics, chemistry, social (default: general)

Date Filtering:
  --recent YEAR           Only papers from this year onwards
  --date-after DATE       Specific start date (YYYY-MM-DD)
  --date-before DATE      Specific end date (YYYY-MM-DD)

Search Mode:
  --pro                   Enable Pro Search (multi-step reasoning)

Output:
  --output FILE           Save results to JSON file
  --verbose               Show detailed information
  --citations-only        Output only citations (for BibTeX generation)

Domain Presets

Preset	Domains	Best For
general	All academic domains	Broad literature search
medical	nih.gov, pubmed.gov, nejm.org, lancet.com, bmj.com	Clinical research
biomedical	nature.com, cell.com, pnas.org, ncbi.nlm.nih.gov	Life sciences
cs	arxiv.org, acm.org, ieee.org	Computer science
physics	arxiv.org, aps.org, iop.org	Physics
chemistry	acs.org, rsc.org, nature.com/nchem	Chemistry
social	.edu, jstor.org, ssrn.com	Social sciences

# Medical/clinical research
python scripts/perplexity_academic.py "diabetes treatment RCT" --domain medical

# Computer science papers
python scripts/perplexity_academic.py "transformer attention mechanisms" --domain cs

# Broad academic search
python scripts/perplexity_academic.py "climate change impacts" --domain general

Programmatic Usage

from scripts.perplexity_academic import academic_search

# Simple search
result = academic_search("What are recent advances in CRISPR?")

# With domain preset and date filter
result = academic_search(
    query="CAR-T therapy clinical outcomes",
    domain_preset="medical",
    recent_year=2023,
    verbose=True
)

# Pro Search for synthesis
result = academic_search(
    query="Compare mRNA vs viral vector vaccines",
    pro=True
)

if result["success"]:
    print(result["answer"])
    print("\nCitations:")
    for citation in result.get("citations", []):
        print(f"  - {citation}")
else:
    print(f"Error: {result['error']}")

Integration with Research Agents

This skill is the primary search tool for research agents:

Agent	Recommended Usage
academic-researcher	Default search tool
literature-reviewer	With --pro for synthesis
fact-checker	Verify claims against scholarly sources
research-synthesizer	Multiple queries, cross-reference

Agent Configuration Example

In agent definitions, reference this skill:

skills: perplexity-academic

Or invoke programmatically:

from scripts.perplexity_academic import academic_search

# In literature-reviewer agent
findings = academic_search(
    query="systematic review methodology guidelines",
    domain_preset="medical",
    recent_year=2020
)

Common Use Cases

Literature Search

# Find recent papers on a topic
python scripts/perplexity_academic.py \
  "machine learning for drug discovery" \
  --recent 2023 \
  --domain biomedical

Citation Discovery

# Find papers to cite
python scripts/perplexity_academic.py \
  "transformer architecture attention mechanisms" \
  --domain cs \
  --citations-only

Research Verification

# Verify a claim against literature
python scripts/perplexity_academic.py \
  "Is intermittent fasting effective for type 2 diabetes?" \
  --domain medical \
  --recent 2020

Systematic Review Support

# Find RCTs in date range
python scripts/perplexity_academic.py \
  "randomized controlled trials SGLT2 inhibitors" \
  --domain medical \
  --date-after 2020-01-01 \
  --date-before 2024-12-31 \
  --pro

Output Format

Results include:

{
  "success": true,
  "query": "Original query",
  "answer": "Synthesized response with inline citations",
  "citations": [
    {
      "number": 1,
      "url": "https://nature.com/...",
      "title": "Paper title if available"
    }
  ],
  "model": "sonar-pro",
  "search_mode": "academic",
  "domain_preset": "general",
  "usage": {
    "input_tokens": 150,
    "output_tokens": 1200
  }
}

Cost Considerations

Academic searches use higher context which increases cost:

Query Type	Approximate Cost
Simple academic query	$0.003-0.005
Pro Search academic	$0.025-0.050
Deep research query	$0.050+

Tips to manage costs:

1. Use specific, focused queries
2. Reserve --pro for complex synthesis needs
3. Use --recent to limit search scope
4. Batch related queries together

Troubleshooting

Academic Mode Not Working

Issue: Results don't prioritize scholarly sources

Cause: Using OpenRouter backend

Solution: Configure direct Perplexity API:

PERPLEXITY_API_KEY=pplx-your-key-here

No Recent Papers Found

Issue: Results are older than expected

Solution:

1. Check --recent year is reasonable (not future)
2. Topic may have limited recent publications
3. Try broader query terms

Missing Citations

Issue: Response lacks source citations

Solution:

1. Ensure using direct Perplexity API
2. Academic mode should always return citations
3. Check API key is valid

Dependencies

Same as perplexity-search:

uv pip install requests python-dotenv

Related Resources

• General search: .claude/skills/perplexity-search/SKILL.md
• Best practices: .claude/docs/perplexity-best-practices.md
• Perplexity API docs: https://docs.perplexity.ai/