pplx-research 1.0.0

Monolithic research CLI powered by Perplexity Sonar API with auto-classification and iterative gap analysis

PPLX Research

A monolithic, versatile research CLI powered by Perplexity's Sonar API. Features auto-classification, iterative gap analysis, multi-perspective synthesis, and comprehensive output formats.

Why This Exists

Other Perplexity wrappers are simple API clients. PPLX Research is a research orchestration layer that adds:

• Auto-classification: Analyzes your query and picks the optimal mode/depth/sources
• Gap analysis: Iteratively identifies and fills knowledge gaps (deep mode)
• Multi-perspective synthesis: Gathers insights from academic, news, forums, docs, and code sources
• Full API coverage: All Perplexity features including search_domain_filter, reasoning_effort, return_images, etc.

Installation

pip install pplx-research

Quick Start

# Set your API key
export PERPLEXITY_API_KEY="pplx-..."

# Basic research
pplx-research "what is quantum computing"

# Let AI pick the best approach
pplx-research "React vs Vue performance" --auto

# Deep academic research
pplx-research "transformer architecture" --mode deep --sources academic --depth 4

# Multi-perspective analysis
pplx-research "microservices pros and cons" --mode synthesis

Modes

Mode	Description	Best For
quick	Single query, fast response	Facts, definitions, current events
deep	Iterative gap analysis	Complex topics, research reports
synthesis	Multi-source perspectives	Comparisons, debates, pros/cons

Options

pplx-research <query> [options]

Modes:
  -m, --mode {quick,deep,synthesis}  Research mode (default: quick)

Scope:
  -s, --site DOMAIN         Constrain to specific domain
  -e, --exclude DOMAIN      Exclude domain (can use multiple times)
  -t, --time-range {hour,day,week,month,year}
                           Time range for results
  -r, --region CODE        Country/region code (US, UK, JP, etc.)
  -l, --language CODE      Language filter (en, de, ja, etc.)
  --sources TYPES          Comma-separated: academic,news,docs,forums,code,all

Output:
  -f, --format {markdown,json,summary,plain}
                           Output format (default: markdown)
  -d, --depth 1-5          Iteration depth (default: 3)

Automation:
  -a, --auto               Auto-classify query and select optimal settings
  -o, --output PATH        Save output to file
  -w, --webhook URL        POST results to webhook on completion
  -q, --quiet              Suppress progress (for piping)

Python API

from pplx_research import ResearchEngine, PerplexitySDK

# High-level research
engine = ResearchEngine(
    query="quantum computing applications",
    mode="deep",
    sources=["academic"],
    depth=4
)
report = engine.run()

# Low-level SDK
sdk = PerplexitySDK(api_key="pplx-...")
result = sdk.chat("Hello", model="sonar-pro")

License

MIT License - see LICENSE for details.