000

click or press any key to skip

← All products

· Nº 02 — Cybersecurity

CERBERUS. — Autonomous AppSec Engine: A security skill book for AI agents: policy guardrails, local audits, Git patches.

Autonomous AppSec Engine

A security skill book for AI agents: policy guardrails, local audits, Git patches.

Request Enterprise Audit WhatsApp Security LeadNPM Package ↗
SYS_02
The Studio · CERBERUS
Program14 GOVERNED SKILLS · LOCAL AST AUDITS

· 001 — In short

AI agents now run shell commands, edit code and touch databases on their own. Cerberus gives them a rule book: every action passes a rule-based policy check first, code is audited on your machine, and fixes arrive as Git patches for a person to approve.

Spec sheet7 specs

Policy skills
14 pre-loaded
Integration
Model Context Protocol (MCP)
Engine
Compiled native binary
Remediation
Git patch & IAM diffs
Data boundary
100% local, zero telemetry
Audit evidence
For SOC 2 / ISO 27001 audits
Distribution
npm & CLI

· 002 — What it does

  1. 01

    Zero-Trust Policy Guardrails

    Autonomous risk classification engine enforcing strict decision gates before any tool execution, shell command, or database migration by AI agents.

  2. 02

    Millisecond AST Code Scanners

    Compiled native binary engine scanning Abstract Syntax Trees (ASTs) and dependency graphs in milliseconds without cloud latency or outbound telemetry.

  3. 03

    Automated AI Remediation

    Generates production-grade Git patch diffs, least-privilege IAM policy rewrites, and security PR reviews ready for one-click human verification.

  4. 04

    Universal MCP Protocol

    Plugs directly into Claude Desktop, Cursor, and enterprise autonomous agents via the Model Context Protocol (MCP) as governed security tools.

· 003 — Install & MCP

Global CLI quickstart
# Install globally via NPM
$ npm install -g @araskova/cerberus
# Run a local AST security audit
$ cerberus audit --strict --auto-patch

Scans codebase, flags credential exposures and dangerous invocations, and generates formatted Git diff patches.

Claude Desktop & Cursor MCP config
{
  "mcpServers": {
    "cerberus": {
      "command": "npx",
      "args": [
        "-y",
        "@araskova/cerberus",
        "mcp"
      ]
    }
  }
}

Gives your local AI agents governed security auditing and policy guardrails via stdio IPC.

· 004 — Next step · Enterprise security deployment

Secure your autonomous AI agents before production.

We deploy CERBERUS on your internal CI/CD pipelines, air-gapped repositories, and local developer environments to ensure zero accidental credential leaks and strict AI tool governance.

Schedule Enterprise Audit

· + — More from Araskova