PT0-003 Sample Questions & Answers

The tester is attempting an adversarial evasion attack. This type of attack involves creating a malicious input (an adversarial example) by making small, often imperceptible, perturbations to a legitimate input. The goal is to cause a deployed and trained machine learning model to misclassify it during the inference phase. In this scenario, the tester would modify the image of the unauthorized person in a specific way to trick the AI model into classifying it as an authorized employee. This is a black-box attack, as the tester interacts with the model's API to observe outputs and iteratively craft the malicious input without needing access to the model's internal workings. Data poisoning occurs during training, while model inversion and membership inference aim to extract information about the model or its data.

Software Composition Analysis (SCA) is the specific methodology designed to identify and manage vulnerabilities in third-party and open-source components. SCA tools scan an application's dependencies (e.g., from package.json, requirements.txt, or container layers) and compare them against a database of known vulnerabilities (CVEs). This allows the CI/CD pipeline to automatically detect and flag insecure dependencies. SAST analyzes first-party source code for coding flaws, DAST tests the running application from the outside, and IAST works at runtime, but only SCA focuses specifically on the security of the software supply chain and third-party libraries.

Since script-src allows 'self', any script loaded from the application's own origin is trusted. A JSONP endpoint on the same domain that reflects user input in a callback function (e.g., /api/jsonp?callback=alert(1)) can be abused to execute arbitrary JavaScript. The injected payload would be , which is permitted by the CSP.

The CSP explicitly trusts scripts from https://trusted.cdn.com. If any JavaScript library hosted on this CDN has a vulnerability (such as an insecure callback, DOM-based XSS sink, or a gadget for prototype pollution), it can be used as a vector to execute arbitrary code. The attacker's payload would call the vulnerable function from the trusted library.

Discovering an active compromise by a real threat actor changes the engagement from a test to a live incident. The consultant's primary responsibility is to immediately stop all testing activities to avoid interfering with forensic evidence or alerting the attacker. The next step is to contact the client via the designated emergency escalation path, which should be defined in the rules of engagement. This allows the client to initiate their incident response plan immediately. Continuing the test or attempting containment could corrupt evidence and is outside the scope of a standard penetration test.

An unquoted service path vulnerability is only exploitable if the user has write permissions in one of the directories in the path. The service path C:\Program Files\Custom App\service.exe will cause Windows to look for C:\Program.exe, then C:\Program Files\Custom.exe, and finally the correct path. The MOST critical validation step is to check if the current user (or a low-privileged user) can write a malicious executable into C:\ or C:\Program Files\. The icacls command is used to check the Access Control Lists (ACLs) on these folders. If write permissions exist, the tester can place a malicious Custom.exe in C:\Program Files\ to achieve privilege escalation when the service restarts. Without write permissions, the vulnerability is not exploitable.

Kerberoasting is an attack that targets service accounts in Active Directory. Any authenticated user can request a Kerberos service ticket (TGS) for any service by its Service Principal Name (SPN). The TGS is encrypted with the service account's NTLM password hash. The primary objective of a Kerberoasting attack is to request these tickets and extract the encrypted portion. The tester can then take these hashes offline and attempt to crack them using tools like Hashcat. Since service accounts often have weak, non-expiring passwords, cracking them is often feasible and can lead to significant privilege escalation.

This technique, known as typosquatting or brandjacking, is a classic software supply chain attack. By publishing a malicious package with a name that is a common misspelling of a legitimate package (common-util instead of common-utils), an attacker can trick developers or automated build systems into downloading and executing malicious code. This code could steal credentials, create backdoors, or compromise the entire CI/CD pipeline. This method directly targets the trust developers place in package repositories and represents a direct attack on the software supply chain. Other supply chain attacks include compromising a legitimate package (dependency confusion) or attacking the build tools themselves.

The Proxmark3 is the industry-standard tool for RFID research and penetration testing. It is specifically designed to interact with both low-frequency (125kHz) and high-frequency (13.56MHz) RFID systems. It can read the data from an original HID Prox card, save it, and then write that data to a blank, cloneable card (like a T5577 card), creating a functional copy. The WiFi Pineapple is for wireless network attacks, the USB Rubber Ducky is a keystroke injection tool, and the HackRF One is a general-purpose Software Defined Radio (SDR).

STRIDE is a threat modeling framework developed by Microsoft that provides a mnemonic for categorizing threats. The acronym stands for Spoofing, Tampering, Repudiation, Information Disclosure, Denial of Service, and Elevation of Privilege. It is designed to help developers and security professionals systematically identify and mitigate potential security threats during the design phase of a project. DREAD is for risk rating, OCTAVE is a risk-based assessment methodology, and PTES is a penetration testing execution standard.

In a serverless architecture, the primary security boundary and attack surface shifts from the network and OS level to the application and identity level. The most critical aspect to test is the Identity and Access Management (IAM) configuration. Each Lambda function has an execution role with specific permissions. Overly permissive roles are a common and high-impact vulnerability, allowing an attacker who compromises one function to pivot and access other AWS services (like S3 buckets or DynamoDB tables) they shouldn't. Therefore, a thorough review of IAM roles and policies is the primary focus for identifying potential data exposure and lateral movement paths. Network scanning is irrelevant due to the shared responsibility model, and while code review is important, the IAM permissions define the blast radius of any code-level vulnerability.