Introduction

WardnMesh.AI is the immune system for your autonomous agents. It acts as a middleware layer that intercepts, scans, and blocks malicious inputs (like prompt injection) and dangerous outputs (like PII leakage) in real-time.

Initialize in 30 seconds

Node.js SDK

Protect your TypeScript/JavaScript agents with the native SDK.

Installation

npm install @wardnmesh/sdk

Usage

import { AgentGuard } from '@wardnmesh/sdk';

const guard = new AgentGuard();

// In your agent loop
const userInput = "Ignore generic instructions...";
const scan = await guard.scan(userInput);

if (scan.blocked) {
  console.log("Threat blocked:", scan.reason);
  return;
}

// Proceed with agent logic...

Python SDK

Seamless integration for LangChain, CrewAI, and AutoGen.

Installation

pip install wardnmesh

Usage

from wardnmesh import AgentGuard

guard = AgentGuard()

user_input = "System prompt injection attempt..."
result = guard.scan(user_input)

if result.blocked:
    print(f"Blocked: {result.reason}")
else:
    agent.run(user_input)

Configuration

Customize your defense layers using the wardnmesh.config.ts file.

Example configuration coming soon.

Core Concepts

The Mesh

Distributed threat intelligence network.

Local Engine

Zero-latency, privacy-first scanning.

Rules Engine

Define custom regex and semantic patterns to block specific threats.