VibeShieldVibeShield
Security scanner for Cursor apps

Free Cursor security scanner

Cursor can help review a generated diff and propose tests. Follow each change through authentication, ownership checks and deployment configuration, then verify the behavior on the host you actually use.

No signup or domain verification. See findings, evidence, check-by-check coverage and a score when checks complete. Checks cover headers, TLS, cookies and public HTML/JS exposure. A free account adds all four tools and AI fix prompts.

See a Cursor scan walkthrough

A 30-second walkthrough of scan options and an example finding.

What we check on Cursor apps

Typical stack: Any stack — AI-written code you deploy yourself

The no-signup scan checks headers, TLS, cookies and public HTML/JavaScript exposure. The checks below describe broader account audits; database access and authenticated workflows also need manual testing. Results include evidence, individual check outcomes and coverage limitations.

Exposed .env and .git files

A misconfigured server can expose sensitive files. Eligible Pro and Business users can opt into bounded hidden-file probes after verifying the hostname. Inspect any affected resource and review the hosting configuration.

Hardcoded keys and tokens

Review whether generated code returns or embeds a private key in browser resources. VibeShield scans shipped JavaScript and page data for supported credential patterns; confirm the key type and provider permissions before acting on the finding.

Permissive CORS

Review CORS findings from inspected public responses. An API that reflects an untrusted Origin while allowing credentials can expose authenticated responses under some cookie and endpoint configurations.

Missing security headers

Review CSP, HSTS and framing protection on the deployed response. Missing headers are configuration risks, not proof of a working exploit. Test the suggested policy with login and integrations before enforcing it.

Educational scenario · manual testing in your own staging app

Review an AI-generated document-export change

  1. Start with the diff: trace the browser request, server identity check, ownership decision and returned document.
  2. Ask for a test that fails before the fix. In staging, the owner should download their document; another user and an anonymous visitor should be denied.
  3. Review the agent's proposed diff and test results, then check production headers and files separately. An agent's explanation is a proposal, not proof.

Ask Cursor for a focused review

Review this document-export diff. Find the server authentication and ownership checks, preserve legitimate sharing, and add a regression test using two synthetic users. Explain what fails before the fix. Do not change unrelated code.

These tests extend beyond a public URL scan. Read the Cursor testing guide and consult OWASP testing guidance.

For public exposure checks, scan your app for free. Account audits need no domain verification; active probes require a verified hostname and an eligible plan.

After reviewing the code, use the secure Cursor deployment checklist to check the published build and production settings.

Found something? Fix it in Cursor.

  • Start with your live URL — no signup for the limited first security scan
  • Signed-in audits cover passive security, accessibility, SEO & AEO, and bundle without domain verification
  • Every issue explained in plain English, ranked by what to fix first
  • Account reports include AI-ready prompts to help your coding tool propose fixes you can review and test

Want to keep checking after changes? Pro adds weekly schedules, email alerts, and active attack checks (after verification). Business adds deploy webhooks, daily schedules, Slack/Discord/ClickUp alerts, and GitHub source scanning with Gitleaks and OSV.dev.

Also built with: Lovable · Bolt · v0 · Replit