Vibe-Code Security Gate by Venkai

    1

    Scans a deployed AI-built app (Lovable, Bolt, Replit, Cursor) for exposed secrets, open Supabase tables, and missing security headers — then returns a SHIP/FIX/BLOCK verdict.

    $29

    Secure checkout via Stripe

    Money-back guarantee

    Converts to your local currency at checkout

    Security scanned

    Works with the AI Agents you already use

    Claude CodeClaude CodeCursorCursorCodex CLICodex CLIMuseMuseDotsDots+21 more

    Vibe-Code Security Gate by Venkai

    Example session with this skill installed

    /vibe-code-security-gate https://juice-shop.herokuapp.com --i-own-this

    (OWASP Juice Shop: the official public app intentionally built vulnerable for security testing — real output, nothing faked.)

    • Read your context and instructions
    • Compiled the vibe-code security gate

    Summary

    Real scan of the OWASP Juice Shop public demo (an app intentionally built vulnerable for security testing) — verdict

    FIX, 6 findings, no fabricated output.

    VERDICT: FIX  (6 constats)
    [HIGH] ADMIN_OPEN: /admin responds 200 with no login required
        -> require a session + a server-side role check
    [HIGH] ADMIN_OPEN: /debug responds 200 with no login required
        -> require a session + a server-side role check
    [HIGH] ADMIN_OPEN: /dashboard/admin responds 200 with no login required
        -> require a session + a server-side role check
    [HIGH] CORS_OPEN: CORS accepts an arbitrary origin
        -> allow-list specific origins
    [MEDIUM] HEADER_MISSING: content-security-policy header missing
        -> add a CSP
    [MEDIUM] HEADER_MISSING: strict-transport-security header missing
        -> add HSTS
    

    Next steps

    • Fix the HIGH findings first (exposed admin routes, open CORS), they're exploitable by anyone with a browser.
    • Add the missing headers (CSP, HSTS).
    • Run the code-review checklist (Step 2 of the skill) for IDOR and write-side RLS, which this black-box scan cannot see.

    Connects securely to your tools. The creator never sees your data.

    About this skill

    AI-generated applications often prioritize functionality over security, leading to recurring vulnerabilities like exposed environment variables or misconfigured database permissions. This skill identifies common security failures in "vibe-coded" stacks including Supabase, Stripe, and modern frontend frameworks before they reach production. It performs a dual-layer assessment: a black-box external scan to detect exposed secrets and misconfigured headers, followed by a targeted code review for logical flaws like IDOR vulnerabilities and missing server-side authorization.

    What it does

    • Secret Detection: Scans JavaScript bundles for leaked Stripe, OpenAI, Anthropic, or AWS keys and service-role credentials.
    • Supabase Audit: Checks for tables with Row Level Security (RLS) disabled or permissive policies allowing unauthorized updates.
    • Configuration Scan: Identifies exposed .env or .git directories and missing security headers like CSP and HSTS.
    • Logic Review: Analyzes API routes for IDOR risks and ensures webhook signatures are verified before processing.
    • Cost Protection: Flags missing rate limits on high-cost LLM endpoints and authentication routes.

    How it works

    1. Target Verification: Confirms ownership or permission to scan the target URL to ensure authorized testing only.
    2. External Scan: Runs a read-only black-box analysis on the deployed URL to find public-facing vulnerabilities.
    3. Internal Review: Audits the repository for specific authorization patterns, secret usage, and input handling.
    4. Verdict Generation: Issues a SHIP, FIX, or BLOCK status based on the severity of identified risks.

    Frameworks & tools

    This tool targets modern web stacks including Supabase, Stripe, Tailwind, and React/Next.js. It uses a specialized Python-based scanner for external audits and supports code analysis for apps built with Lovable, Bolt, Cursor, v0, and Claude Code.

    Why this beats prompting it yourself

    Generic prompts often miss the specific quirks of AI-generated code, such as service-role keys accidentally imported into client components. This skill follows a rigid, read-only protocol that provides exact SQL and code snippets for fixes rather than vague architectural advice.

    Use cases

    • Auditing a Supabase backend for RLS leaks before opening a public beta.
    • Checking for exposed API keys in high-velocity AI-generated frontend builds.
    • Verifying Stripe webhook signature logic before processing live payments.
    • Ensuring .env files and git history are not accessible on a production server.

    Known limitations

    Does not perform write-based testing, brute-force attacks, or load testing. Scans are read-only and cannot detect deep architectural flaws outside the specified focus areas.

    Requires read access to the repository for full audit. Cannot detect vulnerabilities on unreachable sites. Assessment is not a guarantee of total security.

    How to use

    1. 1

      Connect your agent to the Agensi MCP

      Copy the prompt and paste it into Claude, ChatGPT, Cursor or any agent that supports MCP. It sets up the connection for you. Manual setup

    2. 2

      Sign in to Agensi

      Your agent shows a sign-in link. Sign in and click Allow.

    3. 3

      Ask for what you need

      Your agent loads this skill and uses it on your task. It always asks before any payment.

    Reviews

    No reviews yet

    Be one of the first to try it. Every listed skill goes through the trust checks below.

    Security scanned

    Checked by our 8-point scan before listing

    Fresh listing

    Recently published to Agensi

    Money-back guarantee

    Doesn't do what its page promises? You get your money back.

    Trust & safety

    Security scan

    Security scan passed 29 days ago

    Requires

    Git

    Permissions

    Terminal / Shell
    Read Files
    Network Access
    Write Files
    Listed29 days ago

    What's inside

    Frequently Asked Questions

    This skill identifies specific vulnerabilities common in AI-generated code, such as exposed Stripe/OpenAI keys in client bundles, misconfigured Supabase Row Level Security (RLS), and missing webhook signature verification.

    The skill is optimized for apps built using Lovable, Bolt, Cursor, v0, and Claude Code, specifically targeting stacks that utilize Next.js, Supabase, and Stripe.

    Unlike general prompts, this skill follows a rigid, read-only protocol specifically designed to catch service-role leaks and provides exact SQL or code snippets to fix the issues immediately.

    Yes, the skill requires a Target Verification step to confirm ownership or authorized access to the URL before performing its external black-box scan.

    You receive the full dual-layer assessment tool (External Scan and Internal Review) along with the 'Verdict Generation' logic that assigns SHIP, FIX, or BLOCK statuses to your builds.

    The skill performs read-only assessments and does not conduct invasive write-based testing, brute-force attacks, or load testing on your production environment.

    New in Security