Awesome MCP › Other MCP Servers

tanigami/mcp-server-perplexity

⭐ 95 Python added to this list on 2025-05-06 repository created 2024-12-05

mcp-server-perplexity is a small Model Context Protocol server that puts the Perplexity API behind one MCP tool. The tool, ask_perplexity, sends a request to Perplexity for a chat completion and returns the answer together with the citations Perplexity provides, which gives an assistant a way to reach current web-grounded information with sources instead of answering from its own training data alone. The server is deliberately narrow: there is a single tool and no resources or prompts, and all configuration is a PERPLEXITY_API_KEY environment variable set in the MCP client configuration. It is packaged for Python and normally started with uvx, so no local build or checkout is needed; the README shows the mcpServers entry for Claude Desktop on both macOS and Windows and the project is also listed on Smithery for automated installation. The README documents one significant limitation. Perplexity requests can take a long time to process, and the Claude Desktop client may time out before the response arrives. The author notes that this depends on the client supporting long running operations and progress reporting, and says the implementation will be updated to use those features if and when they become available, so users of slower Perplexity models should expect occasional timeouts in the meantime. The project fits users who already hold a Perplexity API key and want research answers with citations available inside a chat client, and it is also a compact reference for what a single tool MCP server written in Python looks like end to end. Because the server only forwards the query and returns the completion, the quality and freshness of results come entirely from the Perplexity model chosen on the account behind the API key.

https://github.com/tanigami/mcp-server-perplexity

mcpmcp-serverpythonperplexitysearchcitationsuvxapi-integration