Perplexity Reasoning MCP Server

A Model Context Protocol (MCP) server that provides reasoning with deepseek R1, web search and citations functionality using Perplexity AI's API. Works with the Anthropic Claude desktop and compatible MCP clients.

Forked from jsonallen/perplexity-mcp and improved in several ways:

1. Deepseek R1, Sonar and Sonar Pro Models
2. You can change the model in the MCP client configuration file
3. Increased the max output tokens to 8K
4. Recency filter adjusted to be included only when the user specifies
5. Citation URLs included in the output

Installation

Prerequisites

1. Python 3.11 or higher - Install from python.org or your system's package manager

   • Windows users: Make sure to check "Add Python to PATH" during installation
   • Linux/macOS users: Most systems come with Python, but you may need to install Python 3.11+

2. UV package manager (requires Python to be installed first):

# Windows
powershell -ExecutionPolicy ByPass -c "irm https://astral.sh/uv/install.ps1 | iex"

# macOS/Linux
curl -LsSf https://astral.sh/uv/install.sh | sh

User Installation (PyPI)

For users who want to use the MCP server:

uv pip install perplexity-mcp-server

Developer Installation (Local)

For developers who want to modify the server:

# Clone the repository
git clone https://github.com/felores/perplexity-mcp-server.git
cd perplexity-mcp-server

# Create and activate virtual environment
uv venv .venv
.\.venv\Scripts\activate  # Windows
source .venv/bin/activate # Unix

# Install in development mode
uv pip install -e .

Configuration

Claude Desktop Setup

1. Get your API key from Perplexity AI

2. Edit Claude's config file:

• Windows: %APPDATA%/Claude/claude_desktop_config.json
• macOS: ~/Library/Application Support/Claude/claude_desktop_config.json

For Users (PyPI Installation)

{
  "perplexity-reasoning": {
    "env": {
      "PERPLEXITY_API_KEY": "your-api-key-here",
      "PERPLEXITY_MODEL": "sonar-reasoning"
    },
    "command": "uv",
    "args": ["run", "perplexity-mcp-server"]
  }
}

For Developers (Local Installation)

{
  "perplexity-reasoning": {
    "env": {
      "PERPLEXITY_API_KEY": "your-api-key-here",
      "PERPLEXITY_MODEL": "sonar-reasoning",
      "PYTHONPATH": "path/to/your/project/src"
    },
    "command": "python",
    "args": ["path/to/your/project/src/perplexity_mcp/run.py"]
  }
}

Available Models

• sonar: Standard model (127k context)
• sonar-pro: Advanced model (200k context)
• sonar-reasoning: Chain of Thought model (127k context)

If no model is specified, defaults to sonar.

Usage

1. Open Claude desktop client
2. Try a prompt like: "Search the web for recent news about AI"
3. Click "Allow for this chat" when prompted

Development

For development instructions, see DEVELOPMENT.md