Ninja research

Ninja Research Skill

Perform deep web research using Perplexity AI with parallel search agents.

Overview

Ninja Researcher leverages Perplexity AI to perform comprehensive research on any topic. It decomposes complex topics into multiple queries, searches in parallel, and aggregates results from diverse sources.

When to Use This Skill

• Researching technical topics or documentation
• Gathering information from multiple perspectives
• Fact-checking claims before presenting them
• Building knowledge bases for complex projects
• Staying current on rapidly evolving technologies

How It Works

Topic -> Query Decomposition -> Parallel Perplexity Searches -> Source Aggregation -> Synthesized Results

Available Tools

researcher_deep_research

Comprehensive multi-query research.

Parameters:

• topic (required): Main research topic
• queries: Specific sub-queries (auto-generated if empty)
• max_sources: Maximum sources to gather (default: 20)
• parallel_agents: Concurrent search agents (default: 4)

researcher_fact_check

Verify claims against web sources.

Parameters:

• claim (required): Statement to verify
• sources: Specific URLs to check against (optional)

Returns: verdict (verified/disputed/uncertain) with confidence score

researcher_summarize_sources

Condense multiple web sources.

Parameters:

• urls (required): List of URLs to summarize
• max_length: Maximum summary length in words

researcher_generate_report

Create structured reports from sources.

Parameters:

• topic: Report subject
• sources: Source documents
• report_type: comprehensive | summary | technical | executive

Example Usage

Deep Research

Use researcher_deep_research:
{
  "topic": "MCP (Model Context Protocol) server development",
  "queries": [
    "MCP protocol specification",
    "Building MCP servers Python",
    "MCP server best practices 2026"
  ],
  "max_sources": 30,
  "parallel_agents": 6
}

Fact Checking

Use researcher_fact_check:
{
  "claim": "Claude 4 was released in December 2025"
}

Best Practices

1. Specific Queries: Break broad topics into focused sub-queries
2. Source Diversity: Use higher max_sources for contentious topics
3. Always Cite: Include source URLs in your final response
4. Cross-Reference: For critical information, verify across multiple sources

Requirements

• Ninja MCP installed: uv tool install ninja-mcp[researcher]
• PERPLEXITY_API_KEY environment variable set