Skip to main content
CERT-In Empanelled Since 2008

Thick Client Penetration Testing

Security testing of desktop applications: the binary, what it stores on the machine, what it holds in memory, what it sends over the network and the server that trusts it.

6,700+
Assessments
1,000+
Clients
150+
Team
2006
Founded

What thick client testing is

A thick client is an application installed on the user’s machine that does much of its work there: a trading terminal, a teller or treasury client, an ERP front end, a point-of-sale or clinical system. Thick client penetration testing attacks that application from the position of someone who owns the machine it runs on.

That position is what separates it from a web test. The attacker holds the binary, the memory and the disk, so every check the client performs can be read, patched or skipped. The test establishes which controls live on the server and which only looked as if they did.

Six layers of a desktop application

A two-tier client and a three-tier client fail in different places. All six layers are examined on both.

The binary

Decompilation of .NET and Java, reverse engineering of native code, and a search for hard-coded credentials, keys, connection strings and hidden functions. Signing and obfuscation are checked.

Local storage

Files, the registry, local databases, configuration and logs: what the application writes, whether it is encrypted and who else on the machine can read it.

Memory and runtime

Secrets left in process memory, runtime instrumentation, DLL search-order hijacking and the patching of client-side checks on role, licence and limits.

Network traffic

Interception of HTTP and proprietary TCP protocols, certificate validation and pinning, and replay or tampering of the messages the server accepts.

The server and database behind it

Whether authorisation is enforced by the server, what a modified client can request, and what a two-tier client’s database account can reach directly.

The host

The installer, the updater and any service the application runs: file and service permissions, the update channel, and escalation from a standard user.

Where the test concentrates

The architecture decides it. These are five common ones.

The client connects straight to a database

The test concentrates on the database account, its privileges and the credentials the client holds to use it.

Configuration and hardening review →

The client is the front end of an API

The API is tested in its own right, with the client’s traffic as the map of what to attack.

API security testing →

You build the application in-house

A secure code review beside the test, so that each finding is traced to the line that causes it.

Secure code review →

The application is published through remote desktop or a kiosk

Breakout testing: whether a user can leave the published application and reach the host or the network behind it.

Network penetration testing →

A vendor supplies the application and you hold no source

Black-box testing of the build you were given, with a report you can put to the vendor.

Penetration testing services →

How the test runs

Static work first, so that the dynamic work knows where to look.

  1. Scoping

    The build under test, the user roles, the environment and the servers it talks to. A test environment with production-like data is the usual choice.

  2. Architecture mapping

    Processes, files, registry keys, network endpoints and trust boundaries are recorded before anything is attacked.

  3. Static analysis

    The binary is taken apart to find what it contains and how it decides what a user may do.

  4. Dynamic analysis

    The running application is instrumented, its traffic intercepted and its checks bypassed, one role at a time.

  5. Server-side testing

    Everything the client was trusted to enforce is requested directly from the server and the database.

What you receive

Security Brigade was founded in 2006 and has 6,700+ assessments delivered for 1,000+ enterprise clients, with 150+ security professionals and offices in Mumbai, London, New York and Singapore.

  • An executive report with the risk picture and priorities.
  • A technical report with reproduction steps, CVSS v4.0 scores and a fix for each finding, split between client and server.
  • A retest of the fixes on the corrected build.
  • A security assessment certificate.

Thick client testing, answered

Request a scoping call
What is a thick client application? +
An application that is installed and run on the user’s own machine and does a large share of its processing there, in contrast to a web application that runs in a browser. Trading terminals, banking teller clients and ERP desktop clients are typical.
How does thick client testing differ from web application testing? +
In a web test the attacker controls the requests. In a thick client test the attacker controls the application itself as well: its code, its memory and its files. A proxy alone misses most of that, so the work adds reverse engineering and runtime instrumentation.
Do you need the source code? +
No. The test works from the installed build, as an attacker would. Where you can share source, a secure code review alongside shortens the work and ties each finding to its cause.
Is a two-tier application riskier than a three-tier one? +
It concentrates the risk. A two-tier client holds a database credential, so anyone who extracts it speaks to the database without the application. A three-tier client moves that question to the server, where authorisation can be enforced.
Can the test run against production? +
It can, and for some applications there is no other environment. The scoping call agrees which actions are safe on live data and which are run only against a copy.

Scope a thick client test

Tell us what the application does and what it connects to. We reply within one business day.