Perplexity
FreePerplexity 官方集成，连接 Perplexity Sonar API，实现实时全网研究。
FreeFree tier
Inputs: textOutputs: text
About Perplexity
The official MCP server implementation for the Perplexity API Platform, providing AI assistants with real-time web search, reasoning, and research capabilities through Sonar models and the Search API.
Key Features
Real-time web search via Perplexity Search API returning ranked results with metadata
Conversational AI with real-time web search using the sonar-pro model
Deep comprehensive research using the sonar-deep-research model
Advanced reasoning and problem-solving using the sonar-reasoning-pro model
Optional strip_thinking parameter to remove think tags and save tokens
Configurable timeout, custom base URL, and log level
Easy integration with Claude Code, Codex, Cursor, Claude Desktop, Kiro, Windsurf, and VS Code
Pros & Cons
Pros
- Official implementation directly from Perplexity AI
- Multiple specialized models for different research needs (search, ask, research, reason)
- Flexible configuration with timeout, base URL, and log level settings
- Integrates seamlessly with popular AI development clients
- Open source and community-supported on GitHub
Cons
- Requires a valid Perplexity API key, which may incur costs
- Dependent on the Perplexity API; no offline or local functionality
- Primarily designed for MCP-compatible clients, limiting standalone use
Best For
Quick fact-checking and everyday searches with conversational AIThorough research and detailed reports using deep research capabilitiesComplex analytical tasks requiring advanced reasoning and problem-solvingGeneral-purpose real-time information retrieval for AI assistants
FAQ
How do I install the Perplexity MCP server?
Install via npx with the command `npx -y @perplexity-ai/mcp-server` and set the PERPLEXITY_API_KEY environment variable. Clients like Claude Code, Cursor, and VS Code can be configured using their respective MCP configuration files.
What tools are available in this MCP server?
Four tools are provided: perplexity_search for direct web search, perplexity_ask for general-purpose conversational AI, perplexity_research for deep comprehensive research, and perplexity_reason for advanced reasoning and problem-solving.
Can I remove the thinking tags from responses?
Yes, you can set the optional parameter `strip_thinking` to true for the perplexity_reason and perplexity_research tools to remove think tags and save context tokens.
What clients are supported?
The server works with Claude Code, Codex, Cursor, Claude Desktop, Kiro, Windsurf, and VS Code, as well as any MCP-compatible client.