Every piece of software has a weak spot, whether we like to admit it or not. If software touches a network, runs someone’s code or lets someone log in, it is exploitable. The real question is not whether your software is vulnerable; it’s where the vulnerability lies.
Software vulnerabilities are no longer only a technical problem. Mandatory vulnerability and incident reporting requirements under the Cyber Resilience Act (CRA) apply from 11 September 20261, and these reporting obligations apply to all in-scope PDEs (Products with Digital Elements: That includes standalone software, apps, firmware and connected hardware sold in the EU), including products already placed on the market before 11 September 2026. Being outside the EU doesn’t get you off the hook: a Swiss manufacturer bears the same obligations as any other manufacturer, and the same applies to UK companies selling into the EU.
A breach of the CRA’s reporting obligations may attract an administrative fine of up to €15 million2. Closer to home, the UK’s voluntary Software Security Code of Practice contains 14 principles that each sit within four themes3. So knowing where your app is vulnerable is now a legal question as well as a technical one.
The UK Government also publishes a free self-assessment form alongside the Software Security Code of Practice, which you can use to check your own software against the Code’s principles and share with customers as evidence of your security practices.
After carefully reviewing various vulnerability reports and assessments, I felt that software professionals, team leads and software organisations could go even further. The following 12 categories, and the points under them, are a direct result of my attempt to help them. I have been in the software industry for over 16 years now, and each of the points below is highly relatable.
Software vulnerabilities: 12 categories
As you go through these categories and points, think about your own software project. At the end, you’ll find a downloadable self-assessment spreadsheet you can use to check your own app for software vulnerabilities.
1. Infrastructure and hosting
- Is connected to a network (Internet or private network)
- Runs on cloud infrastructure (AWS, Azure, GCP or similar)
- Uses a content delivery network
- Runs as background jobs (e.g., cron jobs)
- Runs on Kubernetes or another container platform
- Runs on servers or operating systems that need patching
- Uses default settings from frameworks, servers or cloud services
- Exposes databases, caches or management ports to the internet
- Relies on TLS certificates and domain names that need renewing
- Has DNS records pointing at services you no longer use
2. Software supply chain and build
- Uses external libraries and packages
- Depends on libraries, runtimes or frameworks that have reached end of life
- Accepts new package versions automatically through loose version ranges
- Installs packages that run scripts during installation
- Lets AI coding tools suggest or install packages
- Runs in containers built from public base images
- Is built and deployed through a CI/CD pipeline
- Uses third-party actions, plugins or runners in its pipeline
- Runs pipeline jobs on code from outside contributors
- Publishes its own packages, images or releases to public registries
- Has no up-to-date list (SBOM) of what it’s built from
3. Code and application logic
- Contains code written by AI assistants or copied from forums
- Is written in a memory-unsafe language such as C or C++
- Builds SQL queries, shell commands or HTML by joining strings
- Deserialises objects or data from untrusted sources
- Fetches URLs that users supply (link previews, imports, image URLs)
- Has multi-step workflows such as checkout, sign-up or approvals
- Updates balances, stock or counters under concurrent requests
- Trusts checks made in the browser or mobile app
- Grants access by default when a security check errors or times out
- Shows detailed error messages or stack traces
- Leaves debug modes, test accounts or hidden features on in production
4. APIs and integrations
- Consumes 3rd-party APIs
- Provides an API layer to be consumed by 3rd parties
- Exposes record IDs in URLs or API requests
- Returns more data fields than the screen actually needs
- Offers GraphQL or other flexible query endpoints
- Has old API versions or undocumented endpoints still live
- Receives webhooks from other services
- Holds OAuth tokens that let it act inside customers’ other tools
- Sends emails, SMS or push notifications on users’ behalf
- Uses real-time connections (WebSockets, WebRTC, server-sent events)
- Connects to IoT devices, Bluetooth or other hardware
5. Identity and access
- Stores credentials (username, password, OTP)
- Uses single sign-on
- Lets users sign in with a password alone or SMS codes
- Ships with default or shared admin passwords
- Offers password reset or account recovery flows
- Keeps users signed in with long-lived session cookies or tokens
- Validates JWTs or other signed tokens itself
- Supports multiple user roles and permission levels
- Has an admin panel or back-office dashboard
- Keeps secrets in environment variables, config files or code repositories
- Uses API keys, service accounts or bot identities that never expire
- Reuses the same credentials across environments or services
- Leaves accounts active after staff, contractors or customers leave
6. Data and storage
- Collects personal or sensitive data about users
- Serves multiple customers from a shared database (multi-tenancy)
- Stores files in cloud object storage buckets
- Keeps database backups or snapshots anywhere
- Writes logs that capture user activity or error details
- Caches pages or API responses that contain private data
- Copies customer data into CRM, data warehouse or support tools
- Has staging or test environments holding real data
- Keeps data longer than it needs to
7. Cryptography
- Encrypts data with keys stored next to the data they protect
- Uses outdated algorithms or protocols (MD5, SHA-1, TLS 1.0)
- Uses home-made encryption or custom cryptography code
- Generates session tokens, reset codes or OTPs with weak randomness
- Skips certificate checks when talking to other services
- Sends any traffic unencrypted, even inside your own network
- Can’t rotate keys or swap algorithms without a rewrite
- Protects long-lived data with encryption that isn’t quantum-safe
8. Web and mobile front ends
- Interacts with user inputs
- Provides file uploads/downloads
- Generates previews, thumbnails or PDFs from user content
- Shows content that one user posts to other users (comments, profiles, rich text)
- Changes data through links or forms without anti-forgery tokens
- Loads JavaScript from third-party domains (analytics, chat widgets, CDN libraries)
- Can be embedded inside other websites through iframes
- Accesses browser cache or cookie storage
- Runs as a native app on your mobile
- Ships API keys or secrets inside mobile or front-end code
- Accepts deep links or data from other apps on the device
- Can be decompiled, modified or run on rooted devices
- Updates itself automatically on users’ devices
9. AI and agents
- Takes natural-language prompts from users
- Lets an AI model read emails, documents or web pages it didn’t write
- Lets AI agents call tools, APIs or take actions for users
- Connects AI agents to MCP servers, plugins or other agents
- Uses retrieval (RAG) or a vector database
- Gives an AI model memory that carries over between sessions
- Puts secrets or business rules in system prompts
- Passes AI output straight into code, queries or web pages
- Executes code that an AI model writes
- Pays per token for AI calls that users can trigger at will
- Fine-tunes or downloads models and datasets from outside sources
10. Bots, fraud and abuse
- Has a login page or login API reachable from the internet
- Offers free trials, sign-up bonuses, referrals or discount codes
- Sells limited stock, tickets or booking slots
- Publishes prices, listings or content worth scraping
- Takes payments or holds stored value (cards, credits, wallets)
- Has no limits on how often actions or API calls can be repeated
- Can be knocked offline by floods of traffic
11. Monitoring, response and resilience
- Logs events that nobody is alerted on or reviews
- Has no published way for people to report security bugs
- Takes days or weeks to ship a security fix
- Has never tested restoring from backups
- Has no rehearsed plan for a security incident
- Depends on a single cloud region, provider or supplier to stay up
12. People and third parties
- Has support staff who can reset passwords or MFA on request
- Gives staff standing admin access to production
- Gives agencies, contractors or IT providers access to your systems
- Is developed on laptops that also hold production credentials
- Lets staff paste company data into unapproved AI tools
- Is built by people with no secure coding training
- Is used by humans
Download your free software vulnerabilities self-assessment
Reading a list is one thing. Checking your own app against it is where the real value is. I’ve turned all 12 categories into a free Excel spreadsheet you can use to assess your app. Each category has its own tab, and every point comes with a suggested control and a reference to the framework it comes from (OWASP, NCSC, CISA and others).
Here’s how it works:
- Mark what applies. Go through each tab and mark every point Yes, No or Not sure. Most apps skip whole categories, so this goes quicker than it looks.
- Rate your protection. For every Yes, set the status to Done, Partial or Not started.
- Add an owner and evidence. Note who is responsible and what proves the control is in place. This turns the spreadsheet into something you can show to customers, auditors or investors.
- Check your summary. The Summary tab works out your coverage for each category and gives it a red, amber or green rating, so your weakest areas stand out straight away.
It pairs well with the UK Government’s self-assessment form for the Software Security Code of Practice. Use mine to find your gaps, then use theirs to show how you meet the Code’s principles.
In a nutshell
Every app has software vulnerabilities. The real question is where yours are, not whether you have them or not. Each of the 12 categories in this post, from infrastructure and your build pipeline to AI features and the people who support your users, is a place where an attacker could get in.
The risk is now legal as well as technical: the EU Cyber Resilience Act already requires software makers to report actively exploited vulnerabilities, and the UK’s Software Security Code of Practice sets out what good looks like.
Start by finding out which points apply to your app, check how well each one is protected, and fix the red areas first. The free self-assessment spreadsheet gives you a ready-made way to do all three.
Not sure where your app stands?
Not sure where your app stands? Download the free self-assessment, and if you’d like a second pair of eyes on your results, book a call with me at Remote Winners.
🔗 Check out my other articles on Build vs buy software in the age of agentic AI: 10 questions for SMEs, 5 Engineer Hiring Strategies for Non-Technical Founders, 5 App Testing Strategies for Non-Technical Founders, 5 UX Strategies for Non-Technical Founders, 5 Database Strategies for Non-Technical Founders and 5 Tech Stack Strategies for Non-Technical Founders.
📌 Follow Anjana Silva (LinkedIn) for remote team building and tech tips for remote startups.
