perplexity
Free Perplexity AI MCP server with multi-account pooling, React admin dashboard, zero-cost health monitoring, and Telegram alerts. No API keys needed.
claude mcp add --transport stdio teoobarca-perplexity-mcp uv --directory /path/to/perplexity-mcp run perplexity-mcp
How to use
Perplexity MCP server provides a multi-account pool that rotates among Perplexity accounts, enabling Pro-level access without per-query fees. It includes an admin dashboard for real-time monitoring, token management, and health checks, plus an integrated toolset to perform semantic queries and deep research across multiple sources. Use the provided MCP tools to interact with the server: perplexity_ask lets you query the pool with model choices and source controls, while perplexity_research performs in-depth investigations with extended citations. The server handles round-robin token usage, automatic failover, and zero-cost health checks by querying the Perplexity rate-limit API so your queries aren’t billed in the background.
To use it, install the MCP client (uv) and register the perplexity MCP server in your MCP configuration. The admin panel runs locally (default on http://localhost:8123/admin/) and lets you view quotas per account, monitor health, and export/import token configurations. If you’re using a community or team setup, you can add multiple accounts to the token pool and rely on the built-in hot-reload to apply changes without restarting the server.
How to install
Prerequisites:
- Python 3.10+ installed on your system
- Git installed
- MCP tooling available (uv CLI)
- Clone the Perplexity MCP repository
git clone https://github.com/teoobarca/perplexity-mcp.git
cd perplexity-mcp
- Install or ensure the MCP tooling is available on your PATH
- The MCP tooling (uv) is used to manage and run MCP servers. If you don’t have it, install via your preferred method per MCP docs.
- Prepare the server directory
- Decide where you want the server to reside. The quick start references /path/to/perplexity-mcp.
- Run the MCP server via uv
uv --directory /path/to/perplexity-mcp run perplexity-mcp
- Verify the server is running
- Open http://localhost:8123/admin/ to view the Admin Panel (default port).
- If you’re using a different directory, adjust the path in the command accordingly.
Notes:
- You can add your tokens for Pro access later via the Authentication section of the admin panel or by providing a token_pool_config.json in the project root.
- The server is designed to work with anonymous Perplexity sessions out of the box.
Additional notes
Tips and notes:
- Environment: The admin panel is usually available at http://localhost:8123/admin/. If you change the port, update your firewall rules accordingly.
- Token pool config: Start with anonymous sessions to test, then add tokens for Pro usage via token_pool_config.json. The pool supports multi-account rotation and automatic failover.
- Configuration: You can enable/disable health checks, configure Telegram alerts, and adjust cooldown/backoff strategies in the pool_state.json and related config files.
- Common issues: If the admin panel or MCP tools fail to connect, ensure uv is correctly installed and the directory path in the mcp_config matches your actual server location. Check that the Perplexity cookies (csrf_token and session_token) are correctly injected when using a multi-token pool.
- Troubleshooting: If you encounter a crash, enable verbose logs in the admin panel and verify the atomics/state files (pool_state.json) aren’t corrupted. The server supports hot-reload for token configuration without restarting.
Related MCP Servers
claude-talk-to-figma
A Model Context Protocol (MCP) that allows Claude Desktop and other AI tools (Claude Code, Cursor, Antigravity, etc.) to read, analyze, and modify Figma designs
systemprompt-code-orchestrator
MCP server for orchestrating AI coding agents (Claude Code CLI & Gemini CLI). Features task management, process execution, Git integration, and dynamic resource discovery. Full TypeScript implementation with Docker support and Cloudflare Tunnel integration.
mcp-tasks
A comprehensive and efficient MCP server for task management with multi-format support (Markdown, JSON, YAML)
seedream-image
🚀 PixelMCP | 为你的 Cursor、Claude Code 等集成AI绘画能力,让AI生成的页面不再单调!
voice-status-report
A Model Context Protocol (MCP) server that provides voice status updates using OpenAI's text-to-speech API.
ai-control-framework
Stop shipping non-deployable AI code. Framework with DRS scoring, contract freezing, and 30-min mock timeout. Works with Claude Code, Cursor, Copilot.