Security for software built with AI

Youshippedfast.Let'scheckwhatshippedwithit.

Zerivex scans your AI-built app for leaked keys, open CORS, missing headers and injection bugs, then gives you the exact fix for Next.js, Express or Nginx.

Free plan includes 1 public scan per week • No credit card required
14 check engines
31+ fix recipes
Ownership-verified active scans
zerivex-cli v0.1
Sample output
Sample output
$zerivex scan --target=https://my-app.com --profile=strict --sarif
> Initializing deterministic passive check engines...
> [TLS/HTTPS] Inspecting handshake, cipher suites & certificate validity...
> [HEADERS] Auditing RFC 6797 HSTS, CSP and X-Frame-Options policies...
> [SECRETS] Exposed sensitive file check flagged /.env configuration exposure
> [CORS] Evaluating origin reflection and cookie SameSite attributes...
Summary: 3 policy findings identified across target perimeter.
The Security Lifecycle

Verify. Detect. Defend.

A principled security architecture engineered for modern developers. From domain proof to automated patch verification.

STEP 01Verify

Prove it's your site first.

We never run active intrusive probes against unconfirmed targets. Prove ownership with one simple cryptographic challenge before any active test begins.

  • 4 ownership challenge methods: DNS TXT, HTTP Header, HTML Meta, or File upload
  • Uses tamper-evident tokens: zx_verify_<hex32>
  • Prevents unauthorized security testing against third-party systems
STEP 02Detect

Deterministic checks. No synthetic noise.

Our 14 deterministic engines inspect client bundles, headers, cookies, TLS, CORS, and injection vulnerabilities. If we flag it, we captured the proof.

  • Passive mode audits public endpoints, headers, and AST bundle keys
  • Active mode runs non-destructive SQLi and XSS payloads with strict rate limits
  • All passwords, tokens, cookies, and credit cards are automatically redacted
STEP 03Defend

Multi-framework diffs. One-click re-verify.

Stop copying generic OWASP documentation. Zerivex generates exact before/after patches for Next.js, Express, and Nginx, then re-tests only that specific check.

  • Exact framework code diffs with copyable snippets
  • Targeted "Verify Fix Now" re-runs only the failing check in seconds
  • Continuous monitoring triggers alerts when security scores drop 10+ points
Target Ownership Verification
MODULE 1/3
# DNS TXT Challenge Record
Host: _zerivex-challenge.your-domain.com
Type: TXT
Value: zx_verify_9b8f2c4a7e1d5a8b
Protects third-party systems from unauthorized scans.
Continuous Protection Active100% Deterministic
Automated Problem Discovery

Mistakes AI tools leave behind.

Cursor, Bolt, v0, and Claude write functional code fast—but security boundaries are easily skipped. Zerivex catches the blind spots.

Secrets & Files
CRITICALZX-SEC-*

Publicly readable .env and .git repository metadata

Misconfigured reverse proxies and static servers directly expose environment secrets, database credentials, and Git commit history.

Probes /.env, /.git/HEAD, and /.git/config with content-signature validation.Playbook
Perimeter Hardening
HIGHZX-HDR-*

Missing HSTS & clickjacking protection

Leaves production traffic susceptible to SSL stripping and malicious iframe framing.

Validates RFC 6797 max-age directive and X-Frame-Options DENY/SAMEORIGIN policies.Playbook
API Perimeter
MEDIUMZX-CORS-*

Wildcard CORS origins with credentials enabled

Copy-pasted StackOverflow CORS handlers allow arbitrary websites to make authenticated requests on behalf of your users.

Sends origin reflection probes and validates Access-Control-Allow-Credentials combinations.Playbook
Data Layer
CRITICALZX-ACT-*

Unparameterized queries & SQL injection bugs

AI tools writing raw database strings without parameterization expose user tables to data exfiltration.

Non-destructive active SQLi payloads verify ORM escape guarantees with ownership gating.Playbook
Attack Surface
MEDIUMZX-API-*

Exposed GraphQL introspection & debug routes

Development-only diagnostic endpoints and schema explorers accidentally left publicly accessible in production.

Crawls standard schema query paths and flags schema exposure on production origins.Playbook
AI Code Smells
HIGHZX-AI-*

Leftover debug routes, seed endpoints & GraphQL introspection

Development-only diagnostic endpoints, database seed scripts, and GraphQL schema introspection left accessible in production.

Leftover debug routes, seed endpoints and open GraphQL introspection.Playbook
Deterministic Code Remediation

Exact before & after code patches.

No vague security theory. Zerivex provides ready-to-paste snippets for Next.js, Express, and Nginx, with the CLI verify command to test the fix.

CWE-319A05:2021 Security MisconfigurationTarget file: next.config.mjs
BEFORE: VULNERABLE
// Insecure: No transport security headers configured
const nextConfig = {
  reactStrictMode: true,
};

export default nextConfig;
AFTER: HARDENED PATCH
// Verified: Enforce HSTS with preload and subdomains
const nextConfig = {
  reactStrictMode: true,
  async headers() {
    return [
      {
        source: '/(.*)',
        headers: [
          {
            key: 'Strict-Transport-Security',
            value: 'max-age=63072000; includeSubDomains; preload',
          },
        ],
      },
    ];
  },
};

export default nextConfig;
Targeted CLI Verification:$ zerivex scan --target=https://your-domain.com --check=check_hsts_policy
Verify this on your site

Built with Cursor, Lovable, v0, Bolt or Claude Code? Protect your deployment.

Next.js
React
Express.js
Nginx
Node.js
Fastify
Hono
TypeScript
PostgreSQL
Docker
Next.js
React
Express.js
Nginx
Node.js
Fastify
Hono
TypeScript
PostgreSQL
Docker
Next.js
React
Express.js
Nginx
Node.js
Fastify
Hono
TypeScript
PostgreSQL
Docker
CI/CD Quality Gates

Halt security regressions before deployment.

Integrate Zerivex directly into your deployment pipeline. Fail builds when critical flaws are introduced, view native OASIS SARIF v2.1.0 alerts inside GitHub Code Scanning, and get clear PR comments.

Configurable build-breaker thresholds: block builds on CRITICAL or HIGH findings.
Native OASIS SARIF v2.1.0 output feeds straight into GitHub Security tabs.
Comprehensive markdown summary comment posted to pull requests.
# GitHub Actions — use your ZERIVEX_API_KEY secret
curl --fail --silent --show-error -X POST https://zerivex.tech/api/v1/ci/scan \
  -H "Authorization: Bearer $ZERIVEX_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"targetId":"YOUR_REGISTERED_TARGET_ID","wait":false}'
# Poll the returned pollUrl until scanJob.status is COMPLETED.
# Only then evaluate gate.passed. Failed or incomplete jobs must fail the pipeline.
# Get a complete workflow from Targets → CI/CD in your workspace.
System Infrastructure Probes• Observability and egress firewall status
All engines operational
Diagnostic report
14

Deterministic Check Engines

Zero synthetic false positives across active and passive batteries

31+

Multi-Framework Fix Recipes

Exact before/after patches for Next.js, Express, and Nginx

4

Ownership Challenge Methods

Cryptographic DNS TXT, HTTP header, HTML meta, and file verification

10

Interactive Security Playbooks

Curated developer guides covering AI coding pitfalls and defenses

Verification Protocol

How we verify fixes in seconds.

Traditional scanners force you to wait for a full battery scan just to confirm one patch. Zerivex uses precision targeted re-testing: we re-probe the exact failing check against your target origin, verify the defense is live, and immediately certify the finding as RESOLVED.

Targeted Check Execution

Instead of running all 14 engines, we isolate the specific check rule (e.g. ZX-HDR-001) for instantaneous feedback.

Deterministic Evidence Matching

The probe inspects live responses to confirm headers, token masks, or ORM parameterization are actively blocking attacks.

Tamper-Evident Audit Ledger

Every verification is committed to the SHA-256 monotonic audit vault with immutable timestamps and actor attribution.

Developer Playbooks

Security Academy for AI Builders.

Hands-on vulnerability breakdowns, code examples, interactive quizzes, and CLI verification steps for the top security risks in modern apps.

Explore all 10 guides
AI CODE SMELLS
CWE-1188• BEGINNER

The Developer's Guide to Auditing AI-Generated Web Applications

How AI code assistants systematically introduce security vulnerabilities, and a practical pre-deployment verification workflow.

8 min read
Read guide
API SECURITY
CWE-942• INTERMEDIATE

Fixing Permissive Wildcard CORS with Authenticated Credentials

Why Access-Control-Allow-Origin: * combined with credentials enables complete session hijacking, and how to configure strict origin whitelists.

6 min read
Read guide
AI CODE SMELLS
CWE-798• BEGINNER

Preventing Exposed API Keys & Secrets in AI-Generated Frontend Bundles

Understanding why environment variables with NEXT_PUBLIC_ or VITE_ are baked into client JS, and how to establish strict secret isolation.

5 min read
Read guide
INJECTION DEFENSES
CWE-918• ADVANCED

Defeating Server-Side Request Forgery in Webhooks & Data Fetchers

How unvalidated backend HTTP fetch calls expose internal cloud metadata (169.254.169.254) and private networks, and how to implement socket-level IP pinning.

10 min read
Read guide
API SECURITY
CWE-862• INTERMEDIATE

Securing Next.js Server Actions Against Missing Authorization

Server Actions are publicly accessible POST endpoints by default. Here is how to prevent unauthenticated execution and broken access control.

7 min read
Read guide
INJECTION DEFENSES
CWE-89• INTERMEDIATE

Preventing SQL Injection in Raw Queries and Modern ORMs

Why string concatenation in raw queries and unsafe ORM raw clauses ($queryRawUnsafe) lead to database compromise, and how to use parameterized queries.

8 min read
Read guide
Simple, Transparent Plans

Free proves the problem. Paid fixes it for good.

Start with a free perimeter scan. Upgrade when you need continuous monitoring, automated fix verification, or team governance.

Starter

Essential perimeter security and passive checks for individual developers and AI-built projects.

₹0
What's included
  • 1 Monitored Target
  • 8 Scans per month
  • 1 Team member
  • Perimeter security checks (7 passive engines)
  • Fix verification & remediation code diffs
  • Continuous monitoring & regression alerts
Most chosen

Pro

Full 14-engine battery including active verified probes, crawler reconnaissance, and fix verification.

₹1,499/mo
What's included
  • 3 Monitored Targets
  • 50 Scans per month
  • 3 Team members
  • Full 14 security engines (passive + active verified)
  • Fix verification & remediation code diffs
  • Continuous monitoring & regression alerts

Team

Multi-target fleet protection, CI/CD gates, SARIF export, webhooks, and team collaboration.

₹4,999/mo
What's included
  • 15 Monitored Targets
  • 250 Scans per month
  • 10 Team members
  • Full 14 security engines (passive + active verified)
  • Fix verification & remediation code diffs
  • Continuous monitoring & regression alerts

Enterprise

Custom limits, dedicated support, organization-wide governance, and bespoke compliance reporting.

₹0
What's included
  • Target limits by agreement
  • Scan allowance by agreement
  • Team size by agreement
  • Full 14 security engines (passive + active verified)
  • Fix verification & remediation code diffs
  • Continuous monitoring & regression alerts
View full plan comparison & capabilities matrix
Frequently Asked Questions

Clear answers about security testing.

Everything you need to know about safety, ownership verification, redaction, and plans.

Yes. Our passive scan engines only inspect publicly accessible HTTP response headers, TLS ciphers, and client-side JavaScript bundles without sending intrusive payloads. Active probes are strictly non-destructive, rate-limited, and only execute after you cryptographically prove domain ownership.
Verify. Detect. Defend.

Find what shipped with your AI app.

Launch an instant perimeter scan or speak with our security engineering team for multi-app enterprise coverage.

• No credit card required• 14 deterministic engines• Automatic secret redaction• RFC 9116 security.txt aligned