---
title: "Dependency Security: Third-Party Libraries in a Business Application"
description: "Your application contains code nobody on your team wrote or read. There is usually more of it than your own — and that is fine, as long as somebody looks after it."
canonical: "https://interfase.sk/en/news/dependency-security-third-party-libraries-in-a-business-application"
lang: "en"
last_updated: "2026-11-18"
---

# Dependency Security: Third-Party Libraries in a Business Application

> Your application contains code nobody on your team wrote or read. There is usually more of it than your own — and that is fine, as long as somebody looks after it.

Published: 2026-11-18

- Most of the code in a modern application was not written by your team — and every library pulls in others nobody chose.
- An application that is not updated is not stable, it is frozen; the risk grows with every month without maintenance.
- A new dependency is a decision for years — for small tasks, writing it yourself is often cheaper.
- Without a list of what the application contains, a published vulnerability cannot even be answered with "does this affect us".

Open any modern web application and look at how much foreign code runs inside it. Not a hundred files — often thousands. Your developers wrote the smaller part; the rest are libraries and their own dependencies.

That is not a mistake. Writing everything from scratch would be more expensive and worse. But it does mean the security of the application depends on code nobody in the company has read.

## A dependency pulls in more dependencies

This is the part that surprises people most. You add one library for working with dates, and fifteen more arrive because it needs them. You chose one; sixteen entered the application.

The consequence: **the problem can appear in a package you have never heard of**. That is also why updating is one of the few things that cannot be postponed indefinitely.

## "It did not work, so we left it"

The most common strategy and the worst. An application that has gone three years without updates is not stable — it is frozen. Risk grows in the meantime, because vulnerabilities are published continuously and attacks on known holes are automated.

There is a second effect: the longer it goes unupdated, the more expensive updating becomes. After three years it is no longer maintenance but a project.

The sensible practice is the reverse: **small updates, often**, with tests that catch them. This follows on from what we describe in [CI/CD and safe software deployment](/blog/ci-cd-a-bezpecne-nasadzovanie-softveru) — without automated testing every update is a risk, which is why it gets postponed.

## Knowing what is in the application

When a serious vulnerability appears, the first question is: **does this affect us?** Companies with no overview spend hours or days on it.

The answer should take minutes, and it requires one thing — a maintained list of what the application contains and at which versions, including indirect dependencies. It gets generated automatically at build time and forms part of project handover, which we covered in [handing over a software project](/blog/odovzdanie-softveroveho-projektu-dokumentacia-pristupy-a-znalosti).

## How to choose a new library

Adding a dependency is a decision for years, not for one afternoon. Before adding one, four things are worth looking at:

- **Is the project alive?** When was the last change, how quickly do reported bugs get resolved.
- **Who is behind it?** One person in their spare time is a different risk from a library with several maintainers.
- **How much does it pull in?** A package with three dependencies is a different story from one with eighty.
- **What is the licence?** Some licences have consequences for commercial use, and finding that out retroactively is unpleasant.

And one extra question: **is it actually needed?** For a small task, ten lines of your own code are often cheaper than a dependency you have to maintain for years.

## When a vulnerability appears

Not every one is urgent. What decides is not how severe it is but **whether it is reachable in your application** — whether the affected function is used at all and whether it is exposed externally.

A process that works: verify reachability, update if a fix exists, and if it does not, temporarily restrict access to the affected feature. And record the decision — including the decision not to act, and why.

## The risk is also in the surrounding tooling

A vulnerability does not have to be in the application. It can be in the environment that builds and deploys it — and that usually has access to secrets and to production.

So credentials in CI should be as narrow as possible, and permissions there get reviewed too. The same principle as in [access audits](/blog/audit-pristupov-a-bezpecnostne-logy-vo-firemnej-aplikacii).

## Where to start

With two steps: **find out what the application contains** and **when it was last updated**. If the answer to the second is "we do not know", that is the cheapest security improvement available to the company.

If you need an existing application reviewed or regular maintenance set up, [get in touch](/kontakt). We handle it under [cybersecurity](/riesenia/kyberbezpecnost) and [custom development](/riesenia/vyvoj) — examples in our [case studies](/referencie).


---

Web version: https://interfase.sk/en/news/dependency-security-third-party-libraries-in-a-business-application
llms.txt: https://interfase.sk/llms.txt
