Project Summary

This repository offers a Model Context Protocol (MCP) server designed to integrate Perplexity AI's specialized models, serving as an intelligent research and documentation assistant. It targets developers and researchers seeking an efficient way to leverage advanced AI for complex queries. The project's key benefit is its automatic query complexity detection, routing requests to the most appropriate Perplexity AI model for optimal results, and providing universal search capabilities across all tasks.

How It Works

The server analyzes the complexity of user queries to intelligently route them to one of three Perplexity AI models: Sonar Pro for straightforward searches and quick information retrieval, Sonar Reasoning Pro for multi-step analysis and problem-solving, and Sonar Deep Research for in-depth investigations and detailed reports. This approach ensures that each query is handled by the model best suited for its nature, optimizing performance and accuracy. A significant advantage is its implementation of search capabilities for every task, a feature distinct from the official Perplexity server. Users can also manually override the automatic model selection using the force_model parameter within tool arguments.

Quick Start & Requirements

• Primary Install/Run: Requires Node.js and a Perplexity API key. Setup can be done by cloning the repository and configuring MCP settings, or more conveniently via npx -y perplexity-mcp, which is recommended for macOS users.
• Prerequisites: Node.js (from nodejs.org), Perplexity API key (from perplexity.ai/settings/api).
• Configuration: Involves adding server details to the MCP settings file, specifying the command (node or npx), arguments, and environment variables including PERPLEXITY_API_KEY.
• Links: Perplexity API settings: https://www.perplexity.ai/settings/api.

Highlighted Details

• Intelligent Model Routing: Automatically directs queries to Sonar Pro, Sonar Reasoning Pro, or Sonar Deep Research based on detected complexity.
• Universal Search: Features search capabilities applicable to all tasks, enhancing its utility.
• Tool Integration: Provides examples for using search, reason, and deep_research tools.
• Award-Winning: Secured 1st place at the Cline Hackathon.

Maintenance & Community

No specific details regarding maintenance, contributors, community channels, or roadmap were provided in the README snippet.

Licensing & Compatibility

The README snippet does not specify the project's license type or offer compatibility notes for commercial use or integration with closed-source projects.

Limitations & Caveats

The Deep Research tool may experience timeouts when integrated with certain other tools, such as cline, due to implementation differences; however, the 'reason' tool is noted as a compensating factor for these instances.