Continuous Web Application Security Monitoring
A clean scan on launch day does not describe next week. Every deploy, every new table, and every forgotten test host changes what an attacker can reach. Monitoring repeats the checks and shows you what changed.
GET YOUR BASELINE SCAN
Enter your deployed URL. The first scan shows what is open today. Monitoring then shows what changes.
Continuous web application security monitoring means that you test your live app again and again and compare each result with the last one. You find a new problem soon after a deploy or a setting change makes it, not at the next audit.
An app does not stay the same after launch. Developers push code. An AI editor adds a table. A person changes a setting in a dashboard. A DNS record stays in place after the service behind it is gone. Each change can open a path that the last scan did not see.
This page shows what to monitor, what to alert on, and how to start. To decide if you need monitoring at all, read monitoring vs one-time scanning.
What is the difference between monitoring and a scan?
A scan is one test on one day. Monitoring is the same test, repeated, with a comparison. The comparison turns a long list of findings into a short list of changes.
| One scan | Monitoring | |
|---|---|---|
| Question it answers | Is the app open now? | What changed since the last check? |
| Output | A list of findings | A list of new, fixed, and unchanged findings |
| Timing | When you start it | On a schedule and after important changes |
| Best use | Before launch, after a big fix | After launch, while the app keeps changing |
Monitoring is not uptime checking. An uptime tool shows that the site answers. Security monitoring shows if the answer is safe. A site can be up for a year and leak user data for the same year.
Monitoring is also not a firewall. It finds problems. It does not block an attack as it happens. Keep your other controls in place.
Why does an app become less secure between deploys?
Most new exposure comes from a change that looks normal. These are the common causes.
- A new table or endpoint. AI editors add routes and tables fast. Each new table needs its own access rule. Each new route needs its own ownership check. If one is missing, a user can read data that belongs to another user. See the Supabase RLS checker.
- A secret in client code. A developer gives an environment variable a public prefix. The key then ships in the JavaScript bundle. See the token leak checker.
- A configuration change. A host change or a CDN rule removes a security header or opens CORS too wide. See the security headers checker.
- A dashboard edit. A person turns off a policy to find a bug and forgets to turn it on. No commit exists, so no code review sees it.
- A forgotten host. A staging site or a preview deploy stays online with real data behind it.
- A new advisory. A vulnerability in a package you already ship becomes public after you deploy.
The OWASP Top 10 lists security misconfiguration as a top risk. Misconfiguration is a change problem. It comes from a setting that was right once and is wrong now.
What should you monitor in a production app?
Monitor the exposure that an attacker can reach from outside. Start with the checks below. The VibeEval security checks cover these areas.
| Area | What can change | Alert when |
|---|---|---|
| Database access rules | A new table, a dashboard edit | A table is readable without login |
| Secrets in bundles and pages | A new environment variable | A key appears in a page or bundle |
| Authorization on endpoints | A new route, a refactor | One user can read the object of another user |
| Security headers | A host or CDN change | A header is missing or weaker |
| CORS policy | A debug shortcut | A wildcard origin on a credentialed route |
| TLS settings | A certificate renewal, a host move | A weak protocol or a near expiry |
| Information disclosure | A build setting | A source map, a debug page, or a stack trace is public |
| Third-party scripts | A vendor update | A new tracker or script domain appears |
Put the checks in order of damage. A table open to the public is more urgent than a missing header. Use the severity of each finding to decide the order.
How do scheduled re-checks work in VibeEval?
VibeEval has an opt-in monitoring feature for each target. Use the steps below.
- Add your production URL as a target. A target is one origin, for example
https://app.example.com. - Run a first scan. Choose the DAST Full profile for the first baseline.
- Read the findings and fix the critical and high items.
- Open the completed scan and select Set up monitoring.
- Check the Targets page. It shows the last scan time and the risk counters for each target.
- Switch monitoring off from Targets if you no longer need it.
After you turn it on, VibeEval re-checks the target every week and sends change alerts. Monitoring is an early-access capability. For important releases, do not wait for the weekly check. Start a re-run yourself with Scan Now on the Targets page or Re-run Scan on the scan page.
If your app needs login, give the scan an authentication method. VibeEval supports Bearer token, cookie, basic auth, and form login. Use a test account. Never use an account with real customer data. See running scans for all options.
How do you find what changed?
A scan page shows a trend against the previous scan of the same target. The trend lists the findings that are added and fixed, and the change in risk score. The risk score goes from 0 to 100. See risk score for the formula.
VibeEval keeps one record for each issue on a target. If you run the scan again, it updates the record. It does not make a second copy. This keeps the list short and makes the change easy to read.
Use the scan history to answer three questions.
- What is new? Examine the findings that were added since the last scan. Find the deploy or the edit that caused each one.
- What is fixed? Confirm that the finding is closed. Use Retest on a finding to run only the check that found it.
- What stays open? If a finding stays open for many scans, make a decision. Fix it, or mark it with a status that explains why you accept it.
Scan history is kept for 90 days on Pro and Lifetime plans, with a diff view. Enterprise keeps full history. See plans and limits.
How do you detect a new host or a forgotten app?
A host that you forgot is a risk that no scan of your main app can show. VibeEval scans the targets that you register. It does not search for new hosts for you. You must find the hosts first and then register them.
Use this procedure each month.
- List the hosts you know. Include production, staging, and admin sites.
- Search certificate transparency logs for your domain. Each public certificate shows a host name. See crt.sh for a free search.
- Compare the two lists. Mark each host you did not expect.
- For each unknown host, find the owner. If no one uses it, remove the DNS record.
- Add each live host as a target and run a scan.
- Choose the Include subdomains scope when the main app links to its subdomains.
The guide on discovering forgotten web apps covers subdomain search, preview deploys, and DNS in detail.
Which alerts should you send, and to whom?
Too many alerts teach people to ignore them. Send an interruption only for a real risk to user data. Send the rest in a digest.
| Event | Channel | Reason |
|---|---|---|
| Critical finding | Email and webhook, to the person who can fix it | Data or credentials may be exposed |
| High finding | Fix it in days, not weeks | |
| Scan failed | A failed scan means no coverage | |
| Scan completed | Email or none | Useful after a deploy, noisy otherwise |
| Weekly summary | Weekly digest | Shows the trend for the team |
In VibeEval, you control email under Settings → Notifications. Critical findings, high findings, completed scans, and failed scans are on by default. The weekly security digest is off by default. Turn it on if you want a regular summary.
For a chat tool or a ticket system, add a webhook under Settings → Webhooks. Subscribe to finding.critical for urgent items. Subscribe to finding.new to send each new finding to a ticket queue. Subscribe to finding.statusChanged to follow fixes. VibeEval sends each event as a POST request with a JSON body. See webhooks and notifications.
When an alert arrives, the finding already has evidence. It holds the HTTP request and response that showed the issue. It also has numbered fix steps. Use the Copy for LLM option to give the report to your coding agent.
How do you reduce noise?
Noise is the main reason that monitoring fails. Use these rules.
- Keep the target list short. Monitor the apps that hold user data first.
- Exclude paths that cause trouble. Put routes such as
/logoutor/api/healthin the excluded paths. - Mark accepted findings. Use the finding status so that the team sees a decision, not an open item.
- Wait for review. On a paid plan, an on-call security researcher confirms each automated finding. Findings show a Preliminary badge until review ends. Do not open tickets on preliminary findings. Wait for the review email.
- Send one alert to one owner. An alert that goes to everyone belongs to no one.
What does monitoring not do?
Know the limits. They help you to plan.
- It does not stop an attack. It finds a weakness. A firewall or a rate limit stops traffic.
- It does not replace a person who tests business rules. A pricing error or a broken approval flow needs human thought. See continuous penetration testing for depth testing at each release.
- It does not scan your source code. VibeEval tests the deployed app from outside, as an attacker does.
- It does not cover a host that you did not register as a target.
- It does not fix problems. It shows the problem, the evidence, and the fix steps. A person or a coding agent must make the change.
NIST describes this wider practice as information security continuous monitoring in SP 800-137. The NIST guide covers organization-level programs. For one web app, the weekly re-check and the change alerts above are a good first step.
How is monitoring different from a deploy-time test?
A test in your CI pipeline runs on each deploy and can stop a release. It finds only what is true at that moment. It does not find a dashboard edit or a new advisory. Monitoring runs without a deploy. Use both. See continuous security smoke testing for the CI side.
For a new regression that you fixed, add a permanent check so that it cannot return. The guide on security regression testing shows how.
How do you start?
- Run the free self-audit or a first scan on your production URL.
- Fix the critical and high findings. Use Retest to confirm each fix.
- Select Set up monitoring on the completed scan.
- Open Settings → Notifications and turn on the weekly digest.
- Add a webhook for
finding.critical. - Re-run the scan after each important deploy.
- Read the trend each week and make a decision about each new finding.
If you run many sites, such as an agency, read multi-site security monitoring. If you need a one-off scan of any website, see the website vulnerability scanner. For a developer checklist, use the production security checklist.
Related
Common questions
What is continuous web application security monitoring?
How often does VibeEval re-check a target?
Does VibeEval find new subdomains by itself?
Will a scan harm my production app?
Which alerts does VibeEval send?
Do I still need a pentest if I use monitoring?
Monitor your production app
Run a scan on a target, then turn on monitoring for weekly re-checks and change alerts. 14-day trial, no card.
14-day free trial · Cancel anytime