Web App Pentest Checklist: What Should Be Tested
A good web app pentest checks how your site gathers information, handles logins, controls who can see what, treats user input, manages sessions and resists abuse of your business rules, then proves what is truly exploitable and gives you a written report with fixes.
This post is for owners who are about to buy or scope a web app pentest (a penetration test, where a person tries to break into your site the way an attacker would, with your permission). You will get the phases of a test, a test-area checklist based on OWASP and PTES, and a list of questions to hold up against any quote.
Where this checklist comes from
Two public sources shape most serious web tests:
- The OWASP Web Security Testing Guide (WSTG) is a free, detailed list of what to test in a web app. OWASP is a non-profit that publishes open security standards.
- The Penetration Testing Execution Standard (PTES) describes how a test should run from start to finish. It has seven stages: pre-engagement, intelligence gathering, threat modeling, vulnerability analysis, exploitation, post-exploitation and reporting.
The OWASP Top 10 is a short awareness list of the most common web risks. It is a good summary, but it is not a full test plan. A tester should work from the WSTG, not just the Top 10.
For a small website, those seven PTES stages collapse into four phases you will actually notice: recon, testing, exploitation and reporting.
Phase 1: Recon (finding what is exposed)
Recon means mapping what an outsider can see about your site, mostly with light, low-impact requests, before any deeper testing. A tester looks at:
- Subdomains and hostnames tied to your domain, including forgotten ones like
old.ordev. - The technology in use: CMS (content management system, like WordPress), plugins, frameworks, server software and versions
- Pages, forms, file uploads and API endpoints (the URLs your app uses to send and receive data)
- DNS and email settings such as SPF, DKIM and DMARC, which help stop people from faking email from your domain
- Files that leak information, like backups,
.gitfolders or public config files
You can peek at your own site's headers in a terminal. Only run this on a site you own:
curl -sI https://example.com | grep -iE 'server|x-powered-by|strict-transport|content-security|x-frame|x-content-type|referrer'
A Server or X-Powered-By line showing exact version numbers is a small information leak. You can also run the free security headers tool for a quick read.
Phase 2: Testing (looking for weaknesses)
This is the main work. The tester walks through your app by hand and checks each area below. Automated scanners help, but they miss problems that need judgment. If you want the full contrast, read manual vs automated penetration testing.
The web penetration testing checklist
Use this table as your penetration testing checklist for a web application. It follows the WSTG categories in plain English.
| Test area | What the tester checks | Example finding |
|---|---|---|
| Configuration and deployment | Security headers, TLS (the HTTPS encryption setup), exposed admin pages, default settings, backup files | Admin login page open to the whole internet with no extra protection |
| Identity and registration | Can anyone sign up as anything? Is the account process predictable? | Anyone can register with an admin role by editing a form field |
| Authentication (logins) | Password rules, lockout or rate limiting (capping how many attempts someone can make), password reset, multi-factor login, error messages | Reset links that never expire |
| Authorization (access control) | Can user A see or change user B's data? Can a normal user reach admin features? | Changing /invoice/1001 to /invoice/1002 shows another customer's invoice |
| Session management | Cookie flags, logout, session expiry, session reuse | Session cookie missing Secure and HttpOnly |
| Input validation | SQL injection (tricking the database with crafted input), cross-site scripting (XSS), command injection, file upload abuse | A search box that runs attacker-supplied script in other users' browsers |
| Business logic | Abuse of how your app is meant to work: coupons, prices, order steps, limits | Changing the price in a checkout request to 0.01 |
| Client-side | Browser storage, redirects, third-party scripts, clickjacking | Open redirect on a login link that sends users to a fake page |
| API | Missing login checks, too much data in responses, weak rate limits | An endpoint returns every user's email to any logged-in user |
| Error handling and crypto | Stack traces, verbose errors, weak or missing encryption | Error page reveals database table names |
OWASP ranks broken access control first in its Top 10, and automated tools are poor at spotting it because they do not know who should be allowed to see what. It is also where a human tester earns their fee.
A healthy cookie looks like this:
Set-Cookie: session=abc123; Secure; HttpOnly; SameSite=Lax; Path=/
Secure sends it only over HTTPS. HttpOnly hides it from page scripts. SameSite limits cross-site use.
Phase 3: Exploitation (proving it is real)
A possible weakness is not the same as a real one. In this phase the tester tries, carefully, to confirm each issue and see how far it goes.
For example, an access control flaw is confirmed by actually loading another test user's record. A tester then asks: what is the worst realistic result? That answer sets the severity (how urgent the issue is).
A responsible tester follows rules here:
- They work only inside the agreed scope and with written permission.
- They avoid anything that could knock your site offline, such as heavy load or denial-of-service attacks.
- They use test accounts, not your real customers' data.
- They stop and tell you right away if they find something critical.
PTES also lists post-exploitation, which means seeing what an attacker could reach after getting in. On a small web app, that usually means checking whether one flaw leads to more data or admin access, not roaming through your servers.
Phase 4: Reporting (what you pay for)
The report is the product. A useful one gives you, for every issue:
- A clear title and a severity rating
- Evidence, such as the request, response or screenshot
- The impact, in business terms
- Exact steps to fix it
- A short summary a non-technical person can read
If a report is only a raw scanner export with hundreds of unexplained lines, that is not a pentest report. Also check whether the provider retests after you fix things. Bug Circuit's audits, for example, retest the previous report on every later audit of the same site.
Questions to ask before you accept a quote
Compare any quote against this checklist:
- Scope: Which domains, apps and APIs are included? What about staging?
- Manual or scanner: Does a person test by hand, or is it a scan with a logo on it?
- Logged-in testing: Will they test with test accounts for each user role, such as customer and admin?
- Standards: Do they say they follow the OWASP WSTG or similar?
- Safety: Do they promise no denial-of-service testing and require written authorization?
- Report: Does it include evidence, impact and fix steps?
- Retest: Is a retest included, and for how long?
- Fixes: Are fixes included or sold separately? Is that clear?
- Timing: How long until you get the report?
Price varies widely with scope. Our guide on how much a penetration test costs explains why.
Red flags in a pentest quote
- No named standard or method, just the words 'vulnerability scan'
- A promise you will be '100% secure' afterward (nobody can promise that)
- No sample report or description of what the report contains
- Asking you to send passwords over email or chat
- No written authorization step before testing starts
What a pentest does not cover
A pentest is a snapshot of your site on the days it was tested. It does not watch your site afterward, and it is not a compliance certificate such as SOC 2 or PCI DSS. A small-site audit is also not the same as a multi-week enterprise engagement across many systems. Be wary of anyone who blurs these lines.
If you just want a quick outside look first, try a free website security check. It is passive and does not replace manual testing.
The honest next step
If you run one small website and want a person to test it by hand, Bug Circuit's Essential plan is $49 one-time. It is a small-site manual audit of 1 website, including the API that site itself calls. It is not a multi-week enterprise pentest, and it does not include advanced API testing, extra domains or staging environments. The plan covers 3 manual audits to request within a month. Each audit ends in a full written report with every issue, severity, evidence, impact and fix steps, usually within about 5 business days after your site is verified and the Authorization to Test is signed. We report problems and explain the fix. Fixing is a separate service, quoted per issue. See the pricing page for details and for what the other plans include.
Key takeaways
- A web app pentest should cover recon, testing, exploitation and reporting, built on the OWASP WSTG and PTES.
- Use the test-area table above to check that a quote includes logins, access control, sessions, input handling, business logic and APIs.
- Insist on manual testing with test accounts for each user role. Access control flaws are hard for scanners to find.
- Look for a report with evidence, impact and fix steps, plus a clear retest policy.
- Treat any '100% secure' promise or missing authorization step as a red flag.
A real human security engineer audits your whole site by hand and sends a full report: every issue, its severity, and the exact fix. From $49, with a 14-day money-back guarantee.
See pricing