Perplexity MCP Server logo

Perplexity MCP Server

Connects MCP clients to Perplexity for live search, quick answers, deep research, and reasoning.

Install Perplexity MCP Server

Pick your client, then add it in one click or paste the config.

STDIONode.js
~/Library/Application Support/Claude/claude_desktop_config.json
claude_desktop_config.json
{
  "mcpServers": {
    "perplexity-mcp-server": {
      "command": "npx",
      "args": [
        "-y",
        "@perplexity-ai/mcp-server"
      ],
      "env": {
        "PERPLEXITY_API_KEY": "YOUR_VALUE_HERE",
        "PERPLEXITY_TIMEOUT_MS": "YOUR_VALUE_HERE",
        "PERPLEXITY_BASE_URL": "YOUR_VALUE_HERE",
        "PERPLEXITY_LOG_LEVEL": "YOUR_VALUE_HERE",
        "PERPLEXITY_PROXY": "YOUR_VALUE_HERE",
        "HTTPS_PROXY": "YOUR_VALUE_HERE",
        "HTTP_PROXY": "YOUR_VALUE_HERE",
        "PORT": "YOUR_VALUE_HERE"
      }
    }
  }
}

Settings β†’ Developer β†’ Edit Config, merge this into mcpServers, then fully quit and reopen Claude Desktop.

Values are written into the snippet above so you can paste it as is. They never leave your browser.

Install not yet sandbox-testedβ€” we haven't run this command through our automated sandbox yet. Not a red flag; we're working through the catalog.

Overview

Perplexity MCP Server connects MCP clients to the Perplexity API Platform through live web search and agent capabilities. It exposes perplexity_search, perplexity_ask, perplexity_research, and perplexity_reason for current information retrieval, conversational answers, extended research, and reasoning tasks. Use the hosted Streamable HTTP endpoint for a no-install setup, or run the npm package locally with a Perplexity API key. It fits assistants that need web-grounded answers or research inside supported MCP clients.

Use cases

  • Search the web for current information with ranked results
  • Ask time-sensitive questions backed by live web search
  • Run deeper research tasks from an MCP client
  • Apply Perplexity reasoning to assistant workflows

Key features

  • Real-time Perplexity web search
  • Quick answers through the Agent API
  • Deep research and reasoning tools
  • Remote Streamable HTTP endpoint
  • Local npm and self-hosted HTTP modes
  • Search recency and domain filters

Security & health

What it can reach and how reliably it runs β€” from our own checks, not the maintainer’s claims.

ActiveLast checked Today

Recent health check succeeded.

Supply-chain signal

No high-severity advisories surfaced by our automated scan.

Critical 0High 0Medium 0Low 0

Scanned just now via OSV.dev Β· @perplexity-ai/mcp-server (npm)

Adoption & maintenance

Factual signals from GitHub, npm, and our automated checks β€” not a rating.

GitHub stars2.6kStargazers on the source repository.
npm downloads114kPackage downloads in the last 30 days.
Last commit20d agoMost recent push to the default branch.

How Perplexity MCP Server works

What Perplexity MCP Server does

Perplexity MCP Server gives MCP-compatible assistants access to the Perplexity API Platform. Its tools cover direct web search, conversational questions, deeper research, and reasoning:

  • perplexity_search queries the Perplexity Search API and returns ranked results with available publication and update dates.
  • perplexity_ask handles general questions with real-time web search through the Agent API's fast preset.
  • perplexity_research supports deeper, more comprehensive research tasks.
  • perplexity_reason provides reasoning through the Perplexity platform.

The server belongs in workflows where an assistant needs current web information rather than relying only on its model knowledge.

How it works

Perplexity hosts a remote MCP endpoint at https://api.perplexity.ai/mcp. Clients that support remote MCP servers can connect over Streamable HTTP and send the API key in an Authorization header. This route uses the same tools as the local installation and avoids installing or updating a local package.

For local use, an MCP client starts @perplexity-ai/mcp-server through npx. The process reads PERPLEXITY_API_KEY and calls the Perplexity API. The server can also run in self-hosted HTTP mode, where it serves MCP at /mcp, by building the project and starting its HTTP entry point or by running the supplied Docker image.

Search requests can use a recency filter, restrict results to specified domains, or select Fast Search with search_type: "fast" for lower latency and cost.

Setup and configuration

Create a Perplexity API key in the Perplexity API Portal, then choose a connection method. Cursor, VS Code, Claude Code, Codex, Claude Desktop, Kiro, and Windsurf are shown with configuration paths or commands. Local MCP configurations use the mcpServers structure and start the package with npx -y @perplexity-ai/mcp-server.

The required setting is PERPLEXITY_API_KEY. Optional settings include:

  • PERPLEXITY_TIMEOUT_MS for request timeouts; the default is five minutes.
  • PERPLEXITY_BASE_URL for a custom API base URL.
  • PERPLEXITY_LOG_LEVEL with DEBUG, INFO, WARN, or ERROR.
  • PERPLEXITY_PROXY, followed by HTTPS_PROXY and HTTP_PROXY as fallbacks, for network proxies.

Self-hosted HTTP deployments additionally support PORT, BIND_ADDRESS, ALLOWED_ORIGINS, and ALLOWED_HOSTS. The default HTTP port is 8080, and the default bind address is loopback. Docker can expose the service on port 8080.

Tools and capabilities

perplexity_search is intended for ranked, direct search results and supports recency and domain filters. perplexity_ask is suited to quick, general-purpose questions backed by live search. perplexity_research is the option for broader research tasks, while perplexity_reason is intended for reasoning requests. Together, these tools let an MCP client select between retrieval, quick answers, research, and reasoning.

Limitations and notes

Using the hosted endpoint requires an MCP client with remote-server support; clients without it can use the local process instead. API access requires a valid Perplexity API key. When self-hosting HTTP mode, binding to 0.0.0.0 exposes the service on all interfaces, so configure allowed origins and hosts for the deployment. Proxy URLs must use the https:// scheme. API usage, latency, and costs depend on the selected Perplexity API capabilities, including Fast Search.

Reviews

No reviews yet β€” be the first to share how it worked for you.

Frequently Asked Questions about Perplexity MCP Server

Perplexity MCP Server is an MCP server that connects MCP clients to the Perplexity API Platform. It exposes perplexity_search for web results, perplexity_ask for quick answers, perplexity_research for deep research, and perplexity_reason for reasoning, using a Perplexity API key.

Install it locally by configuring your MCP client to run npx -y @perplexity-ai/mcp-server and setting PERPLEXITY_API_KEY. Alternatively, connect a remote-capable client to https://api.perplexity.ai/mcp with a Bearer API key.

Yes, Perplexity MCP Server works with Cursor. Cursor can connect to the hosted endpoint, or you can add a local mcpServers entry that runs the npm package and supplies PERPLEXITY_API_KEY.

Yes, configure PERPLEXITY_PROXY for the preferred proxy setting, or use HTTPS_PROXY or HTTP_PROXY. The server checks those variables in that order, and proxy URLs must use https://.

Related MCP servers

Add a badge to your READMEShow it’s listed β€” live SVG, Markdown or HTML.

AllMCPs Directory Badge

Full Badge Customizer

Showcase your server listing on GitHub or your project documentation. Embed this dynamic SVG badge to highlight official listing status and live engagement.

Badge Style:
Live Dynamic SVG PreviewPerplexity MCP Server AllMCPs Directory Badge
Markdown (GitHub README)
[![AllMCPs](https://allmcps.com/api/badge/perplexity-mcp-server?style=directory)](https://allmcps.com/mcp/perplexity-mcp-server)
HTML Embed
<a href="https://allmcps.com/mcp/perplexity-mcp-server"><img src="https://allmcps.com/api/badge/perplexity-mcp-server?style=directory" alt="Perplexity MCP Server on AllMCPs" /></a>