Built for vibe-coded & AI-generated apps

Don't trust the vibe. Test it.

LARP is the security testing layer for vibe-coded applications. Scan what your AI tool built for broken auth, exposed secrets, and insecure configs before your users do.

3

Scan depths: baseline, full, API

Full log

Every finding the scanner printed

What a report comes back as

Rules checked against your live app

Fail0
Warn11
Pass56

Scanner

OWASP ZAP

Run for you on GitHub Actions

Baseline scan takes

5–10 min

PageSpeed Insights

Metrics

67
Performance
88
Accessibility
96
Best practices
100
SEO

How it works

From vibe-coded to security-tested in three steps

01

Paste your URL

Sign in with GitHub or Google and give LARP the address of your deployed app. Nothing to install, no infra to stand up.

02

Pick how deep to go

Baseline crawls and reads responses without touching your app. Full scan attacks it with real payloads. API scan works from an OpenAPI, SOAP, or GraphQL definition.

03

Read the whole log

OWASP ZAP runs on GitHub Actions and streams its output straight back to you — every rule it passed, warned on, or failed, kept in your scan history.

Security tests

Everything vibe-coded apps tend to get wrong

AI coding tools optimize for working demos, not hardened production apps. LARP runs OWASP ZAP against your deployed app and reports what it finds — passively by default, or with real attack traffic when you ask for it.

Baseline

Security headers

Missing CSP, HSTS, anti-clickjacking, X-Content-Type-Options, and Permissions-Policy.

Baseline

Cookie handling

Cookies set without HttpOnly, Secure, or SameSite, and cookies scoped too loosely.

Baseline

Information disclosure

Debug errors, stack traces, suspicious comments, and sensitive values left in URLs.

Baseline

Vulnerable dependencies

Known-vulnerable JavaScript libraries shipped in your bundle, via Retire.js.

Baseline

Cross-domain configuration

Over-permissive CORS and scripts pulled in from domains you do not control.

Baseline

Transport & caching

Mixed content, HTTP-to-HTTPS form posts, and responses cached when they should not be.

Full scan

Cross-site scripting

Reflected and stored XSS, probed with real payloads against your live app.

Full scan

Injection

SQL, NoSQL, command injection, and path traversal, actively attempted.

API scan

API endpoints

Every route described by your OpenAPI, SOAP, or GraphQL definition, exercised in turn.

Ship fast. Break safely.

Sign in with GitHub or Google and get three scans a day, free. Or run a Lighthouse speed test right now without an account.

Pick your depth

3 a day
BaselineNever attacks
Full scanReal payloads
API scanFrom your spec