Skip to main content

Scan any website
for security issues.

For the developer who wants to know whether what they just shipped has a hole in it, and for whoever is responsible for every domain the company owns, including the four nobody remembers. Paste a URL and the request goes out from our servers, not your browser: 905+ checks come back with the response evidence behind each one, a finding ID that does not change between runs, and a config snippet you can paste straight into Nginx, Caddy, Express, or Next.js.

The free header checkers give you a letter grade and stop. This gives you the rule that fired, the bytes it fired on, and the fix. The whole engine is GPL-3.0 in the public repo, so you can read every check or run the lot yourself.

checks
905+
categories
18
per scan
<3s
licensed
GPL-3.0

Six lines out of 905+ checks. The same request grades the certificate and the TLS handshake, reads SPF, DKIM and DMARC, checks every cookie attribute, and looks through what the page ships for exposed secrets.

This is a finding, in full

Nothing is summarised away. You get the response we saw, the reason it is a problem, and the config change that closes it.

Finding IDs are stable. Put hsts-missing in a PR description, a ticket, or a CI gate and it still resolves to the same check next month.

See a full report

Missing HTTP Strict Transport Security

No Strict-Transport-Security header on the response. Browsers get no instruction to stay on HTTPS, so the first request of every session is downgradeable.

Category
headers
Check ID
hsts-missing
Severity
high

Evidence

> GET / HTTP/1.1
> Host: example.com

< HTTP/1.1 200 OK
< Content-Type: text/html
< X-Frame-Options: SAMEORIGIN
< (no Strict-Transport-Security header)

Fix

# Nginx
add_header Strict-Transport-Security \
  "max-age=63072000; includeSubDomains; preload" always;

# Express
res.setHeader('Strict-Transport-Security',
  'max-age=63072000; includeSubDomains');

What actually happens when you hit scan

The request leaves our servers, not your browser. No session cookies, no VPN, no corporate proxy in the path. What comes back is what an unauthenticated stranger on the internet sees, which is usually the thing you actually wanted to know.

From there the response fans out to 18 independent modules at once. One reads the headers. One finishes the TLS handshake and looks at the certificate chain. One resolves DNS and checks SPF, DMARC and CAA. One walks the HTML and the linked JavaScript looking for keys people did not mean to ship. None of them wait for each other.

No model decides your severity. The same URL produces the same finding IDs and the same ratings on every run, which is the only reason a diff between two scans means anything.

Every finding carries the raw evidence it was derived from, so you can check our work rather than take it on faith. If you disagree with a call, the check ID tells you exactly which rule fired, and the rule is in the public repo.

Most scans are done in under three seconds. Targets with slow DNS or a long certificate chain take longer, and the report tells you how long yours took.

The part that matters

Run it twice, get the same answer twice

Every check is deterministic. Same URL, same finding IDs, same severities, same evidence. There is no model in the detection path deciding today that something is medium when yesterday it was low.

That is what makes the diff between Tuesday and Friday readable, and it is what lets you fail a build on a specific ID without chasing false alarms every other week.

Nothing to install

No agent, no sidecar container, no YAML. The scanner runs on our infrastructure against a URL you give it. A browser extension exists if you want one, but nothing has to be installed.

Fixes, not lectures

Each finding ships with a config snippet for Nginx, Caddy, Express, and Next.js. Copy it, deploy, rescan, watch it disappear.

The report an auditor asks for

Export the same scan as SARIF for the GitHub Security tab, or as a PCI DSS, SOC 2, ISO 27001, ASVS, HIPAA and GDPR crosswalk. Both on the free tier, and the crosswalk says which findings it could not map rather than forcing them into a control.

The UI has no special API

The dashboard calls /api/v3/scan the same way your pipeline does. If you can see it in the interface, you can get it as JSON.

The whole check list

Every check is gated to the URL types it applies to, so a static site is not marked down for missing an API rate-limit header.

905+ checks in 18 categories

Browse every check
Headers149

CSP, HSTS, X-Frame-Options, referrer policy, permissions policy.

SSL15

Certificate chain, signature algorithm, issuer, expiry date.

TLS22

Protocol version, cipher suite, ALPN negotiation, OCSP stapling.

Content156

XSS sinks, reflected parameters, open redirects, mixed content.

Cookies38

Secure, HttpOnly, SameSite, scope, __Host- and __Secure- prefixes.

Config25

Server banner, framework fingerprint, exposed debug endpoints.

Info disclosure58

Source maps, .env files, .git exposure, stack traces in errors.

DNS28

SPF, DMARC, DKIM, DNSSEC, CAA records, dangling CNAMEs.

Email37

MX records, SMTP TLS, SPF alignment, spoofing surface area.

API64

CORS policy, rate-limit headers, GraphQL introspection, OpenAPI.

Code120

Inline JS patterns, vulnerable library versions, leaked tokens.

Secrets62

AWS keys, Stripe, GitHub, OpenAI, generic high-entropy strings.

AI code37

Placeholder auth, disabled TLS verification, weak crypto: the gaps generated code ships with.

Client-side26

DOM XSS sinks, postMessage origin checks, unsafe-inline CSP, prototype pollution.

Supply chain49

Exposed lock files, dependency manifests, build artifacts, source maps.

Host validation13

Open redirects, SSRF through the Host header, subdomain takeover markers.

Reputation3

Google Web Risk lookup for known malware, phishing, and unwanted-software listings.

Active probing5

Opt-in only: submits real values through discovered forms and flags XSS, SQL and template injection, open redirects and CORS from the response itself. No blind or time-based detection.

Who actually runs this

Developers

You want to know whether the thing you just shipped has an obvious hole in it, before someone else tells you.

Scan a single page during review, quote the finding ID in the PR, paste the fix into your server config, and wire the same endpoint into CI with one curl call.

Security teams

You are responsible for every property the company owns, including the four nobody remembers.

Bulk scan a batch of URLs in one request, 5 at a time on the free tier and up to 100 on the top plan, schedule recurring runs with webhook alerts, and hand out share links that a non-engineer can read without a walkthrough.

Platform and DevOps

You want a security gate in the pipeline that does not become its own piece of infrastructure to maintain.

One REST endpoint, bearer token, plain JSON out. Nothing to deploy next to your app, rate limits that survive a monorepo, and a self-host path if the SaaS is a non-starter.

request.sh
curl -X POST https://vulnradar.dev/api/v3/scan \
  -H "Authorization: Bearer $VULNRADAR_TOKEN" \
  -H "Content-Type: application/json" \
  -d '{
    "url": "https://example.com",
    "scanners": ["headers", "tls", "content"]
  }'
response.json
{
  "url": "https://example.com",
  "scannedAt": "2026-06-23T19:02:11Z",
  "duration": 2841,
  "checksRun": 907,
  "summary": {
    "high": 1, "medium": 3, "low": 2, "info": 4, "total": 10
  },
  "findings": [
    {
      "id": "hsts-missing",
      "title": "Missing HTTP Strict Transport Security",
      "severity": "high",
      "category": "headers",
      "evidence": "Header 'Strict-Transport-Security' not present.",
      "fixSteps": [
        "Add Strict-Transport-Security to all HTTPS responses.",
        "Set max-age to at least 31536000."
      ]
    }
  ]
}

One endpoint. Bearer token. JSON out.

The engine behind the dashboard sits at /api/v3/scan. Same IDs, same severities, same evidence you see in the interface. Drop it in a GitHub Action or a cron job and stop opening the site to check.

Auth
Bearer tokens with scan:write, scan:read, and scan:delete scopes per key, encrypted at rest
Bulk
/api/v3/scan/bulk takes 5 URLs per request on the free tier, 100 on the top plan
Webhooks
Fire on completion into Slack, Discord, or your own handler
Reference
Every field and every finding ID is documented
Playground
Send live calls in the browser and copy each one as curl, Python, JavaScript, Go, PHP, Java, Ruby, or C#

It is all in the open

Every check, every API route, and every screen in the dashboard is in the public repo under GPL-3.0. There is no closed detection engine bolted on behind the login. What runs in the SaaS is what you can run yourself.

Self-hosting is one Dockerfile and a Postgres connection string, and CI checks that on every build: it builds the production image, boots it against an empty database, and fails if the schema does not come up. If you disagree with a severity rating, open the check, argue with it, and send a pull request.

License
GPL-3.0
Stack
Next.js, PostgreSQL, TypeScript
Detection engine
In the repo, not a binary
Issue tracker
Public, on GitHub
Our own hardening checklist
SECURITY-POSTURE.md

What we never keep

The login you scan with. Authenticated scanning takes a username, token, or cookie in the scan request itself. It is held in memory for that one scan and is never written to a database, a log line, or the saved result. What gets recorded is that the scan was authenticated and by which method, never the material. There is no credential vault here to breach, which is also the honest trade-off: an authenticated scan is something you run, not something you leave on a schedule.

Analytics about you. No product telemetry and no third-party trackers, in the hosted app or in a copy you run yourself.

Questions people ask first

What is VulnRadar?
VulnRadar is an open-source web vulnerability scanner. You paste a URL and it runs 905+ deterministic checks across 18 categories from our servers: security headers, TLS and certificates, cookie flags, DNS and email records, exposed secrets, server misconfiguration, information disclosure, client-side and supply-chain risks, and the gaps common in AI-generated code. An opt-in active-probing mode additionally tests for SQL injection, cross-site scripting (XSS), template injection, and command injection. Every finding comes back with a severity, the response evidence behind it, and a concrete fix. It is GPL-3.0 licensed and can be self-hosted.
Is VulnRadar open source?
Yes. VulnRadar is licensed under GPL-3.0 and the entire detection engine is in the public repository, so you can read exactly what every check looks for and run the whole scanner yourself with no plan limits.
Is it free to use?
Yes. The free tier runs 25 scans a day with no card, and keeps every result, with no history limit. Paid plans raise the daily limit. The detection engine is identical on every plan.
Do I need to install anything?
No. Scans run from our servers against a URL you provide: no agent, no sidecar container, and nothing to deploy alongside your application. There is an optional browser extension for Chrome and Firefox if you want to scan the page you are looking at, but nothing needs installing to use the product.
How long does a scan take?
Under 3 seconds for a standard scan. Checks run in parallel, and the same URL produces the same finding IDs every time, so two runs can be diffed against each other.
Can I self-host it?
Yes. The whole thing is GPL-3.0 and runs on your own infrastructure. The self-hosting guide covers deployment, the database, and configuration.
What does the scanner actually check?
Security headers, TLS and certificate configuration, cookie flags, DNS and email records, exposed secrets, server misconfiguration, information disclosure, client-side risks, supply chain exposure, and the security gaps common in AI-generated code: 18 categories in total. With active probing enabled it also submits real values through discovered forms to test for SQL injection, reflected XSS, server-side template injection, OS command injection, and open redirects.
Does VulnRadar test for SQL injection and XSS?
Yes. Alongside the passive checks, VulnRadar has an opt-in active-probing mode that submits real payloads through the forms and parameters it discovers and confirms reflected cross-site scripting (XSS), error-based SQL injection, server-side template injection (SSTI), OS command injection, and open redirects. Active probing is off by default and only runs against targets you explicitly authorize, so a standard scan stays passive and safe.
Is there an API?
Yes. Everything the interface can do is available over REST with bearer token auth, plus webhooks for delivering results and scheduled scans for recurring checks.

Paste a URL and see what comes back

25 scans a day on the free tier, no card. The demo needs no account at all.

Scan Any Website for Security Issues | VulnRadar