Penetration testing following OWASP
A controlled attack on your application, following OWASP
By Brian Sahertian · Published on
A pentest, or penetration test, is a controlled attack on your website, web application or API. A tester uses the methods of a real attacker to try to get in, see data or gain rights they should not have. Where a cybersecurity scan looks broadly at whether the basics are in place, a pentest goes deep into one application. ClickSmash tests following the OWASP standards, so you know what was tested, and the outcome can be compared with earlier and later tests.
In short
- A pentest goes deep into one web application or API, and is mostly manual work.
- We test following the OWASP Top 10, the API Security Top 10, ASVS and WSTG.
- Four steps: scope, testing, report and retest.
- We can also fix the findings and keep the application secure afterwards.
What is OWASP?
OWASP, the Open Worldwide Application Security Project, is an open community that publishes freely available standards for software security. Developers, testers and auditors around the world use them as a common language. Four of them matter for a pentest:
- OWASP Top 10: the ten most important risks for web applications. The 2025 edition is the current one.
- OWASP API Security Top 10: the same approach for APIs, with risks such as requesting someone else's objects by changing a number in the URL.
- ASVS (Application Security Verification Standard): a list of concrete security requirements per area, such as login, sessions, access rights and input validation, in three levels. It sets how deep the testing goes.
- WSTG (Web Security Testing Guide): the handbook of test methods. It describes per topic how to test, so a pentest can be repeated.
The OWASP Top 10 of 2025
These are the risks we cover in every web application pentest:
- A01 Broken Access Control: users can reach data or functions that are not meant for them, for example another customer's invoices.
- A02 Security Misconfiguration: unsafe default settings, exposed admin areas, overly detailed error messages.
- A03 Software Supply Chain Failures: vulnerable or compromised packages, libraries and build steps.
- A04 Cryptographic Failures: data stored or sent without encryption, or with weak encryption.
- A05 Injection: input that is executed as code, such as SQL injection and cross-site scripting.
- A06 Insecure Design: flaws in the design itself, such as a password reset that can be guessed.
- A07 Authentication Failures: weak login, no protection against password guessing, sessions that do not expire.
- A08 Software or Data Integrity Failures: updates, files or data that are not checked for tampering.
- A09 Security Logging and Alerting Failures: an attack is not recorded or does not trigger an alert.
- A10 Mishandling of Exceptional Conditions: the application handles errors and unexpected situations unsafely, and then for example allows something that is normally blocked.
Web application pentest
For a web application, portal or web shop we test what a visitor without an account can do, and what a logged-in user can do with the rights they have. Among other things we look at:
- Login, password reset, two-factor authentication and sessions.
- Access rights between roles and between customers: can customer A reach customer B's data?
- Input in forms, search fields, uploads and URLs.
- The business logic: discounts, orders, payments and approvals that can be carried out in the wrong order or more often than intended.
- Server and application settings, security headers, and what the application gives away about itself, such as software versions in headers like X-Powered-By and Server, and details in error messages.
API pentest
APIs are often the least examined part of an application, while most data passes through them. A mobile app, an integration with an accounting package or your own frontend talks to an API, and that API must check for itself who may do what. We test following the OWASP API Security Top 10, including:
- Whether you can request or change someone else's objects by changing a number.
- Whether tokens and API keys are checked properly and expire.
- Whether an API returns more data than the app shows.
- Whether there are limits on the number of requests, so the API cannot be scraped or brought down.
- Whether integrations with other systems, such as webhooks, can be abused.
If your application has an AI feature, such as a chatbot or an assistant that works with your data, we test that too, following the OWASP Top 10 for LLM applications: can a clever question make the AI ignore its instructions, show other people's data or carry out actions that are not intended?
How does a pentest work?
1. Intake and scope
We agree on what is tested: which application, which environment, which roles and which APIs. Together we choose the ASVS level and the approach: black box (without prior knowledge, like an outside attacker), grey box (with test accounts for each role, the approach we usually recommend) or white box (with access to the code as well). We put everything in a written assignment, with the period, the contacts and what is off limits, such as testing on production data. Based on the scope you get a quote in advance.
2. Execution
We prefer to test on a test or acceptance environment that matches production. Automated tools and AI help to map the application and find known vulnerabilities. Most of it is manual work by a tester: a scanner does not find flaws in access rights, business logic or combinations of small mistakes. If we find something serious, we report it straight away, not only in the report.
3. Report
You get a report with a plain-language summary for management, and a technical part for the developers. For each finding it says what we found, how we found it, how serious it is, which OWASP category it falls under and how to fix it. We go through the report with you.
4. Fixes and retest
Your own developers fix the findings, or we do. Then we test the findings again, and you get an updated report that shows what has been fixed. You can show that report to customers, an insurer or an auditor.
When does a pentest make sense?
- Before a new application or a major new feature goes live.
- When a customer, insurer or tender asks for it, for example because that customer falls under NIS2.
- When your application processes personal, medical or financial data. The GDPR requires appropriate security, and regular testing is part of that. Read more about GDPR and AI privacy.
- When you take over an application from another supplier, or when it was built quickly with AI.
- Periodically, for example every year, because code, packages and attack techniques change.
Why ClickSmash
We build and manage web applications and APIs ourselves, in Laravel, Django and WordPress among others, so we know where developers make the mistakes an attacker looks for. That means it does not stop at a report: we can also fix the findings, and keep the application secure afterwards with hosting and management. Read more about our approach at IT security. Want a broad view of where you stand first? Start with the cybersecurity scan for SMEs.
Sources
Want your application tested?
Tell us which application or API you want tested. After a short intake you get a proposal for the scope and a quote.
Request a pentestOr call 020 717 3360
Frequently asked questions
What is the difference between a pentest and a security scan?
A scan looks broadly at whether the basics are in place, such as two-factor authentication, updates, e-mail security and backups, and uses many automated checks. A pentest goes deep into one application or API and is mostly manual work: a tester really tries to get in, also through access rights and business logic.
What is OWASP?
The Open Worldwide Application Security Project, an open community that publishes freely available standards for software security. Well-known ones are the OWASP Top 10, the API Security Top 10, the verification standard ASVS and the testing guide WSTG.
What does a pentest cost?
That depends on the size: the number of screens, roles and API endpoints, and the ASVS level. After the intake you get a quote. If you first want a broader picture at a fixed price, a cybersecurity scan starts from about € 1,000.
How long does a pentest take?
For an average web application or API usually a few days of testing, plus the report. We agree on the exact duration when we set the scope.
Do you test on our live environment?
Preferably not. We test on a test or acceptance environment that matches production. If it has to be production, we agree in advance what is and is not allowed, and at what times.
Can a pentest damage my application?
We test in a controlled way and within the agreed limits, and do not run attacks that can bring systems down unless explicitly agreed. That is why we prefer to test on a separate environment, with a recent backup.
What is a retest?
After the findings have been fixed, we test them again. You get an updated report showing what has been fixed, which you can show to customers, an insurer or an auditor.
Do you also test AI features and chatbots?
Yes, following the OWASP Top 10 for LLM applications. Among other things we check whether a clever question makes the AI ignore its instructions, show other people's data or carry out actions that are not intended.
Can you also fix the findings?
Yes. We build and manage web applications and APIs ourselves, so we can fix the findings and keep the application secure afterwards. Your own developer or supplier can do that too; we then do the retest.
Related projects
Our other services
Websites, apps and software
Websites, web shops, apps, portals and integrations your users enjoy working with. Built faster with AI.
Hosting and management
Updates, security and backups taken care of, also for applications built by someone else or by AI. You no longer have to think about it.
IT security
Advice and extra protection for your software and websites, now that hacking attempts and AI bots are on the rise.
AI in your business
Automating processes with AI: save time and FTEs or handle more customers with the same team. Set up securely.
Software and AI advice
What is allowed, what is possible, what it costs and what it returns. On paper, so you can decide.