Humane Proxy

Overview

Humane Proxy is an AI safety middleware server built on the Model Context Protocol that detects self-harm and criminal intent in LLM prompts. Any MCP-compatible AI assistant can call Humane Proxy's tools directly, letting the model decide when to use them during conversations or agent runs.

The server runs locally as a Python package via uvx humane-proxy, communicating over stdio on your machine. It requires an API key for authentication. Version v0.6.0 is published under the Apache-2.0 license with source code available at Vishisht16/Humane-Proxy on GitHub, where the project has 29 stars.

Humane Proxy is categorized as AI & Agents on mcp.site and integrates with all major MCP clients including Claude Desktop, Claude Code, Cursor, VS Code, Windsurf, and Zed. If you maintain this project, claim your listing to verify ownership, earn the Verified badge, and keep information current.

Installation

// claude_desktop_config.json
{
  "mcpServers": {
    "humane_proxy": {
      "command": "uvx",
      "args": [
        "humane-proxy"
      ]
    }
  }
}