perplexity

cotdp's avatarfrom cotdp

AI-powered web search and reasoning using Perplexity API through the scraper MCP server. Invoked when searching the web with AI synthesis, conducting research with citations, performing complex reasoning tasks, or answering questions requiring real-time information. Provides two tools for different query types.

⭐4stars🔀0forks📁View on GitHub🕐Updated Dec 23, 2025

When & Why to Use This Skill

The Perplexity skill integrates advanced AI-powered web search and deep reasoning capabilities into Claude. By leveraging the Perplexity API, it enables real-time information retrieval, multi-source synthesis, and complex analytical reasoning with verifiable citations. This skill is essential for overcoming knowledge cut-offs and performing high-stakes research that requires up-to-the-minute accuracy and logical depth.

Use Cases

  • Technical Research: Fetching the latest documentation, best practices, and implementation patterns for rapidly evolving frameworks like Next.js 15 or React 19.
  • Architecture & Strategy: Conducting deep-dive comparisons between technical architectures (e.g., Monorepo vs. Polyrepo) or cloud platforms, including trade-off analysis and performance evaluations.
  • Real-time Market Intelligence: Tracking current events, latest AI model releases, and industry news to provide contextually relevant and timely insights.
  • Security & Compliance Analysis: Investigating the security implications of different coding patterns, such as JWT storage methods, using the reasoning-pro model for multi-step vulnerability assessment.
  • Fact-Checking & Citation: Verifying claims and gathering evidence from multiple web sources with direct URL citations to ensure data integrity in reports and documentation.
nameperplexity
descriptionAI-powered web search and reasoning using Perplexity API through the scraper MCP server. Invoked when searching the web with AI synthesis, conducting research with citations, performing complex reasoning tasks, or answering questions requiring real-time information. Provides two tools for different query types.

Perplexity AI Skill

AI-powered web search and reasoning capabilities via the scraper MCP server.

When to Use This Skill

  • Searching the web with AI-synthesized answers
  • Research tasks requiring multiple sources with citations
  • Complex reasoning and multi-step analysis
  • Questions about current events or real-time information
  • Comparing options or analyzing trade-offs

Available Tools

Tool Model Best For
mcp__scraper__perplexity sonar General queries, quick searches, factual lookups
mcp__scraper__perplexity_reason sonar-reasoning-pro Complex analysis, comparisons, multi-step reasoning

Tool Usage

For straightforward queries and information lookup:

mcp__scraper__perplexity(
    messages=[
        {"role": "user", "content": "What are the latest features in Next.js 15?"}
    ],
    model="sonar",
    temperature=0.3,
    max_tokens=4000
)

Response includes:

  • content: AI-synthesized answer with citation markers [1], [2], etc.
  • citations: Array of source URLs
  • model: Model used
  • usage: Token statistics

2. Complex Reasoning

For analytical queries requiring deep thinking:

mcp__scraper__perplexity_reason(
    query="Compare React Server Components vs traditional SSR approaches. Consider performance, developer experience, and migration complexity.",
    temperature=0.3,
    max_tokens=4000
)

Use cases:

  • Technology comparisons
  • Architecture decisions
  • Trade-off analysis
  • Multi-factor evaluations

Parameters

Parameter Type Default Description
messages array required Conversation history (perplexity only)
query string required Question to reason about (perplexity_reason only)
model string "sonar" "sonar" or "sonar-pro"
temperature number 0.3 Creativity 0-2 (lower = more focused)
max_tokens integer 4000 Maximum response length

Conversation Format

The perplexity tool accepts conversation history:

mcp__scraper__perplexity(
    messages=[
        {"role": "system", "content": "You are a technical advisor."},
        {"role": "user", "content": "What database should I use for a real-time chat app?"},
        {"role": "assistant", "content": "For real-time chat, consider..."},
        {"role": "user", "content": "How does Redis compare to PostgreSQL for this use case?"}
    ]
)

Example Use Cases

Technology Research

mcp__scraper__perplexity(
    messages=[
        {"role": "user", "content": "What are the best practices for implementing authentication in Next.js 15 App Router?"}
    ]
)

Current Events

mcp__scraper__perplexity(
    messages=[
        {"role": "user", "content": "What are the latest AI model releases in 2025?"}
    ]
)

Product Comparison

mcp__scraper__perplexity_reason(
    query="Compare Vercel, Netlify, and AWS Amplify for deploying Next.js applications. Consider pricing, features, performance, and developer experience."
)

Architecture Decisions

mcp__scraper__perplexity_reason(
    query="Should I use a monorepo or polyrepo structure for a microservices architecture? Consider team size of 15 developers, 8 services, and CI/CD complexity."
)

Code Pattern Research

mcp__scraper__perplexity(
    messages=[
        {"role": "user", "content": "What is the recommended way to handle form validation in React 19 with server actions?"}
    ]
)

Security Research

mcp__scraper__perplexity_reason(
    query="Analyze the security implications of storing JWT tokens in localStorage vs httpOnly cookies. Consider XSS, CSRF, and token refresh patterns."
)

Best Practices

Choose the Right Tool

  • perplexity: Factual queries, lookups, simple questions
  • perplexity_reason: Comparisons, trade-offs, complex analysis

Temperature Settings

# Factual accuracy (recommended for most queries)
temperature=0.3

# More creative/exploratory responses
temperature=0.7

# Maximum creativity (use sparingly)
temperature=1.0

Working with Citations

Responses include numbered citations [1], [2], etc. matching URLs in the citations array:

{
    "content": "Next.js 15 introduces several features [1] including improved caching [2]...",
    "citations": [
        "https://nextjs.org/blog/next-15",
        "https://nextjs.org/docs/caching"
    ]
}

Iterative Research

Build on previous answers:

# Initial query
mcp__scraper__perplexity(messages=[
    {"role": "user", "content": "What is Zustand?"}
])

# Follow-up with context
mcp__scraper__perplexity(messages=[
    {"role": "user", "content": "What is Zustand?"},
    {"role": "assistant", "content": "Zustand is a state management library..."},
    {"role": "user", "content": "How does it compare to Redux Toolkit for a medium-sized application?"}
])

Requirements

The Perplexity tools require PERPLEXITY_API_KEY environment variable to be set on the scraper MCP server. Without it, these tools will not be available.