VerditNxtGen MCP Suite

Add cryptographic compliance logging to any MCP server in one line.

VerditNxtGen runs a daily Ghost Swarm that sandbox-tests 190+ open-source MCP tools in Docker, scores them for deployment reliability, detects OWASP vulnerabilities at runtime, and generates HMAC-SHA256 signed manifests for SOC2 compliance.

This repository contains the client-side SDKs and proxy that connect your MCP infrastructure to the VerditNxtGen compliance ledger.

---

Packages

Package	Language	Install
verditnxtgen-mcp-middleware	TypeScript	npm install verditnxtgen-mcp-middleware
verditnxtgen-mcp	Python	pip install verditnxtgen-mcp
Sidecar Proxy	Docker	docker pull ghcr.io/shopfarnow/verditnxtgen-sidecar

Get an API key at verditnxtgen.com/connect-mcp — free tier, no credit card.

---

TypeScript — one line

import { McpServer } from "@modelcontextprotocol/sdk/server/mcp.js";
import { withVerditLedger } from "verditnxtgen-mcp-middleware";

const server = withVerditLedger(
  new McpServer({ name: "my-server", version: "1.0" }),
  { apiKey: process.env.VERDIT_API_KEY }
);
// Every tool call is now HMAC-signed and logged to the compliance ledger.
// Zero latency impact — telemetry posts fire-and-forget in the background.

Python — one decorator

from mcp.server.fastmcp import FastMCP
from verditnxtgen_mcp import verdit_ledger

mcp = FastMCP("my-server")

@mcp.tool()
@verdit_ledger(api_key="YOUR_VERDIT_API_KEY")
async def query_database(sql: str) -> str:
    ...

Docker Sidecar — zero code changes

services:
  verdit-sidecar:
    image: ghcr.io/shopfarnow/verditnxtgen-sidecar:latest
    environment:
      VERDIT_API_KEY: "${VERDIT_API_KEY}"
      DOWNSTREAM_URL: "http://your-mcp-tool:3000"
    ports:
      - "8080:8080"
  # Point your gateway at verdit-sidecar:8080 instead of your tool directly

---

What you get

After logging starts, the VerditNxtGen Intelligence Dashboard shows:

• Compliance Ledger: HMAC-SHA256 signed manifest for every tool — chain-of-custody proof for SOC2 Type II audits
• OWASP scan results: LLM01 prompt injection, LLM06 PII boundary, LLM08 vector weakness detection from live sandbox runs
• pass@k curves: Statistical deployment reliability (pass@1, pass@3, pass@5) across 700+ sandbox runs
• Cost intelligence: Compute cost per run across resource profiles

---

Links

• Live platform
• Intelligence Dashboard
• Score Methodology
• Get API key
• Main repo (private — proprietary Ghost Swarm engine)

---

License

MIT — see individual package directories.