Mcp

Overview

Mcp is a Model Context Protocol (MCP) server: Detect prompt injection, jailbreaks, and code injection in untrusted text before it reaches an LLM. In practice that means any MCP-compatible AI assistant can call Mcp's tools directly — the model decides when to use them in a conversation or agent run.

It runs locally: the npm package @safeprompt.dev/mcp speaks MCP over stdio on your machine. Configuration happens through environment variables — you'll need SAFEPROMPT_API_KEY. The current release is v0.1.0, published under the MIT license, with source at ianreboot/safeprompt on GitHub.

Mcp 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 Mcp for safeprompt.dev, claim this listing to verify ownership, earn the Verified badge, and keep the details current.

Installation

// claude_desktop_config.json
{
  "mcpServers": {
    "mcp": {
      "command": "npx",
      "args": [
        "-y",
        "@safeprompt.dev/mcp"
      ],
      "env": {
        "SAFEPROMPT_API_KEY": "<safeprompt-api-key>"
      }
    }
  }
}

Requires SAFEPROMPT_API_KEY