Agent Safety

Verified Integration

Author: @spawning•Category: ServerApplication

GitHub Repository

JSON-RPC 2.0

Protocol Standard

Sub-second

Execution Latency

Active

Operational Status

Client Configuration

— Connect Agent Safety to Claude Desktop or Cursor in seconds

{
  "mcpServers": {
    "agent-safety": {
      "command": "npx",
      "args": [
        "-y",
        "@modelcontextprotocol/server-agent-safety"
      ],
      "env": {}
    }
  }
}

Paste into ~/Library/Application Support/Claude/claude_desktop_config.json (macOS) or %APPDATA%\Claude\claude_desktop_config.json (Windows).

Architecture & Capabilities

System Overview

Provides a unified safety layer for AI agents, integrating cost management, prompt injection scanning, and decision tracing capabilities through an MCP server.

Indexed Date

7/21/2026

License

Open Source

Protocol Layer

stdio / SSE RPC

Frequently Asked Questions

Architecture and operational details for Agent Safety

Agent Safety is a unified safety layer for AI agents, delivered as an MCP server. It integrates essential tools for cost management, prompt injection scanning, and decision tracing to secure and optimize your AI applications.

Related MCP Servers

Browse all servers →

@configuring

0xArchive

Provides comprehensive historical and real-time crypto market data, including orderbooks, trades, candles, and more, from Hyperliquid, HIP-3, and Lighter.xyz.

MCP Server5 FAQs

Learn more

@spawning

0xGasless

Facilitates gasless blockchain transactions and interactions directly from Claude AI conversations.

MCP Server4 FAQs

Learn more

@spawning

0xGasless

Bridges Claude AI with blockchain networks, enabling gasless transactions, swaps, and transfers directly from natural language conversations.

MCP Server5 FAQs

Learn more