Perplexity MCP Server

The Perplexity AI MCP Server bridges the gap between Claude‑style AI assistants and Perplexity AI’s powerful search engine. By exposing a lightweight MCP endpoint, developers can invoke sophisticated web‑search queries directly from their AI workflows without leaving the context of a conversation. The server supports all official Perplexity models—sonar, sonar-pro, sonar-reasoning, and the high‑capacity sonar-reasoning‑pro—allowing assistants to choose the right balance of speed, accuracy, and context length for each task.

At its core, the server implements a single tool, , that accepts a query string, an optional model selector, and a result count. The tool forwards these parameters to Perplexity’s API, retrieves structured search results, and returns them in a JSON format that AI assistants can parse and present. Because the server runs as an MCP service, any client that speaks the protocol—Claude, Claude‑AI, or custom agents—can call it with a simple JSON payload and receive instant search results.

Key features include:

• Model flexibility: Switch between lightweight or reasoning‑heavy models on the fly, giving developers fine control over latency versus depth of understanding.
• Configurable result count: Limit or expand the number of search hits (1–10) to fit the assistant’s output budget.
• Robust error handling: The server logs detailed diagnostics to , simplifying troubleshooting in production environments.
• Inspector compatibility: The service is fully compatible with MCP Inspector, enabling developers to test and debug tool calls without deploying a full agent.

Typical use cases span from knowledge‑base enrichment (fetching up‑to‑date facts during a conversation) to dynamic data retrieval in customer support bots, and even as a backend for research assistants that need real‑time web evidence. In multi‑tool pipelines, the search tool can feed results into downstream reasoning or summarization modules, creating a seamless chain of AI operations.

Because the server is written in TypeScript and leverages a minimal dependency set, it can be deployed quickly on any Node.js‑capable host. Its straightforward environment variable configuration (just the Perplexity API key) and clear logging make it a drop‑in solution for developers looking to augment their AI assistants with reliable, high‑quality search capabilities.