Perplexity MCP Zerver is an MCP server that connects AI assistants to Perplexity.ai using browser automation, bypassing the need for dedicated API keys. Designed for software developers, technical researchers, and knowledge workers, it enables LLMs to run live web searches, retrieve structured documentation, analyze code for deprecated patterns, and discover development APIs directly through Perplexity's web interface. It runs locally using Bun and Node.js, storing ongoing conversation state and research histories inside a local SQLite database for context preservation across sessions. In addition to standard search queries, the server includes dedicated extraction capabilities for web articles and GitHub repositories. Users with Perplexity Pro accounts can authenticate via an interactive browser login step to route queries through advanced models while preserving persistent local profiles. By abstracting the web UI through standard Model Context Protocol tools, it gives external MCP clients access to cited web information and deep technical answers without requiring usage-based API billing or remote third-party data tracking.
Category: Browser & Web Automation
Tags: perplexity, puppeteer, search, web scraping
bash git clone https://github.com/wysh3/perplexity-mcp-zerver.git cd perplexity-mcp-zerver bun install bun run build 3. (Optional) Run bun run login to authenticate with your Perplexity Pro account via email. 4. Add the server to your MCP client configuration file: json { "mcpServers": { "perplexity-server": { "command": "bun", "args": ["/absolute/path/to/build/main.js"], "timeout": 300 } } } Replace /absolute/path/to/build/main.js with the actual path to your built file.Part of MCP Servers
Perplexity MCP Zerver is an open source Model Context Protocol server that automates interactions with Perplexity.ai using Puppeteer. It allows MCP-compatible AI clients to perform web searches, retrieve developer documentation, evaluate code patterns, and inspect URLs without needing an official Perplexity API key.
No, Perplexity MCP Zerver does not require a Perplexity API key. The server uses browser automation through Puppeteer to send queries through the web interface, making it free to run. Users can optionally log into an existing Perplexity Pro account to utilize premium reasoning models.
The server exposes several tools to MCP clients, including search for web research, get_documentation for technical reference lookups, find_apis for discovering development APIs, check_deprecated_code for analyzing code patterns, extract_url_content for web and GitHub parsing, and chat_perplexity for persistent contextual conversations.
Perplexity MCP Zerver persists conversation history locally using an embedded SQLite database. This design maintains multi-turn context across queries without transmitting conversation logs to third-party tracking services or requiring remote session databases.
Yes, Perplexity MCP Zerver is open source software distributed under the GNU General Public License v3.0. The full source code and licensing details are available in the project GitHub repository.