Humanizer is a Model Context Protocol (MCP) server: Transform AI-generated text into natural, human-sounding content that bypasses AI detectors. In practice that means any MCP-compatible AI assistant can call Humanizer's tools directly — the model decides when to use them in a conversation or agent run.
You can run Humanizer two ways: connect to the hosted endpoint at https://api.gpthuman.ai/mcp with no local install, or run it on your own machine via the npm package @gpthuman/mcp-server. Configuration happens through environment variables — you'll need GPTHUMAN_API_KEY. The current release is v1.0.4, published under the Apache-2.0 license, with source at GPTHuman-ai/mcp-server on GitHub.
Humanizer is listed under AI & Agents on mcp.site and works with any MCP client — Claude Desktop, Claude Code, Cursor, VS Code, Windsurf, Zed and the rest of the ecosystem — using the install snippets below. If you maintain Humanizer for GPTHuman-ai, claim this listing to verify ownership, earn the Verified badge, and keep the details current.
