Perplexity Chat

The Perplexity MCP Server provides a Python-based interface to the Perplexity API, offering tools for querying responses, maintaining chat history, and managing conversations. It supports model configuration via environment variables and stores chat data locally. Built with Python and setuptools, it's designed for integration with development environments.

The MCP Server is desined to mimick how users interact with the Perplexity Chat on their browser by allowing your models to ask questions, continue conversations, and list all your chats.

- ask_perplexity: Request expert programming assistance through Perplexity. Focuses on coding solutions, error debugging, and technical explanations. Returns responses with source citations and alternative suggestions.
- chat_perplexity: Maintains ongoing conversations with Perplexity AI. Creates new chats or continues existing ones with full history context. Returns chat ID for future continuation.
- list_chats_perplexity: Lists all available chat conversations with Perplexity AI. Returns chat IDs, titles, and creation dates (displayed in relative time format, e.g., "5 minutes ago", "2 days ago"). Results are paginated with 50 chats per page.
- read_chat_perplexity: Retrieves the complete conversation history for a specific chat. Returns the full chat history with all messages and their timestamps. No API calls are made to Perplexity - this only reads from local storage.

-

Model Configuration via Environment Variable:Allows you to specify the Perplexity model using thePERPLEXITY_MODELenvironment variable for flexible model selection.

You can also specifyPERPLEXITY_MODEL_ASKandPERPLEXITY_MODEL_CHATto use different models for theask_perplexityandchat_perplexitytools, respectively.

These will overridePERPLEXITY_MODEL. You can check which models are available on thePerplexitydocumentation.

- Python 3.10 or higher
- uvxpackage manager installed

Note: Installation instructions for uvx are availablehere.

"mcpServers": { "mcp-perplexity": { "command": "uvx", "args": ["mcp-perplexity"], "env": { "PERPLEXITY_API_KEY": "your-api-key", "PERPLEXITY_MODEL": "sonar-pro", "DB_PATH": "chats.db" } } }

Configure the MCP Perplexity server using the following environment variables:

npx -y @smithery/cli@latest run @daniel-lxs/mcp-perplexity --config "{\"perplexityApiKey\":\"pplx-abc\",\"perplexityModel\":\"sonar-pro\"}"

Theask_perplexitytool is used for specific questions, this tool doesn't maintain a chat history, every request is a new chat.

The tool will return a response from Perplexity AI using thePERPLEXITY_MODEL_ASKmodel if specified, otherwise it will use thePERPLEXITY_MODELmodel.

Thechat_perplexitytool is used for ongoing conversations, this tool maintains a chat history. A chat is identified by a chat ID, this ID is returned by the tool when a new chat is created. Chat IDs look like this:wild-horse-12.

This tool is useful for debugging, research, and any other task that requires a chat history.

The tool will return a response from Perplexity AI using thePERPLEXITY_MODEL_CHATmodel if specified, otherwise it will use thePERPLEXITY_MODELmodel.

Lists all available chat conversations. It returns a paginated list of chats, showing the chat ID, title, and creation time (in relative format). You can specify the page number using thepageargument (defaults to 1, with 50 chats per page).

Retrieves the complete conversation history for a givenchat_id. This tool returns all messages in the chat, including timestamps and roles (user or assistant). This tool doesnotmake any API calls to Perplexity; it only reads from the local database.

The MCP Perplexity server now includes a web interface for easier interaction and management of chats.

- Interactive chat interface
- Chat history management
- Real-time message display

WhenWEB_UI_ENABLEDis set totrue, the web UI will be available athttp://WEB_UI_HOST:WEB_UI_PORT.

By default, this ishttp://127.0.0.1:8050.

This project uses setuptools for development and builds. To get started:

python -m venv .venv source .venv/bin/activate # On Linux/macOS # or .venv\Scripts\activate # On Windows

Install the project in editable mode with all dependencies:

The virtual environment will contain all required dependencies for development.

This project is open to contributions. Please see theCONTRIBUTING.mdfile for more information.

This project is licensed under the MIT License. See theLICENSEfile for details.

An AI-powered email intelligence platform that integrates with Gmail and OpenAI. It can be run as a CLI tool or deployed on AWS Lambda for enhanced capabilities.