In short
- Start with three pages: the OWASP Top 10:2025 for what goes wrong, the OWASP Cheat Sheet Series for how to fix it, and your platform's production checklist for the settings that matter on your stack.
- The OWASP Top 10 is an awareness document, in OWASP's own words. The list of requirements you can verify one by one is the ASVS, at version 5.0.0 since 30 May 2025.
- PortSwigger's Web Security Academy is free and comes with interactive labs. Practise attacks there or on a local OWASP Juice Shop, and on nothing you do not own.
- Free tools answer narrow questions well: response headers, secrets in a repository, known flaws in dependencies. None of them tells you whether a stranger can read your tables.
- Reading tests nothing. After the list, run the checks on the live app, and run them again after each deploy.
On this page
- Where to start if you have one hour
- The standards and references to bookmark
- Free training with real labs
- Checklists from the platforms themselves
- Free and open-source tools, and the one question each answers
- Guides for a specific job: data, search and LLM teams
- What reading does not replace
- Questions
- Sources
Where to start if you have one hour
Most lists of web app security resources run to hundreds of links. You do not need hundreds. If you have shipped an app and have one hour, four stops cover the ground, in this order.
- 1
Read the risk list (10 minutes)
Open the OWASP Top 10:2025. It has ten categories, from A01 Broken Access Control to A10 Mishandling of Exceptional Conditions. Read A01, A02 and A07 in full: in apps built by prompting, these three carry most leaks. Our reading of the list for AI-built apps maps each category to a check.
- 2
Run your platform's checklist (20 minutes)
Supabase publishes a production security checklist, Firebase documents its Security Rules, and Next.js has a guide on data security. The vendor knows which of its settings leave data exposed.
- 3
Run three free checks on the live app (15 minutes)
Response headers with the security headers checker, keys in your JavaScript with the API key leak checker, and row level security with the Supabase RLS checker.
- 4
Bookmark the fixes (5 minutes)
The OWASP Cheat Sheet Series has one page per topic, such as authentication, session management, HTTP headers and file upload. Open the page that matches each thing you found.
The rest of this page is the longer list, sorted by the job you are doing: learning, configuring a platform, testing, or working in a neighbouring field such as data, search or LLM cost.
The standards and references to bookmark
| Resource | What it is | Use it when |
|---|---|---|
| OWASP Top 10:2025 | The ten most critical risk categories for web applications. OWASP calls it a standard awareness document. | You need a shared vocabulary, or a customer asks whether you cover the OWASP Top 10. |
| OWASP ASVS 5.0.0 | The Application Security Verification Standard: numbered requirements for designing, developing and testing web applications. Version 5.0.0 was published on 30 May 2025. | You want a list you can verify line by line, or a contract asks for one. |
| OWASP Cheat Sheet Series | Concise pages on specific application security topics, written by practitioners. | You know the problem and want the accepted fix. |
| OWASP Web Security Testing Guide | OWASP's testing resource for web application developers and security professionals. The current release is v4.2, and version 5.0 is in development. | You plan a manual test, or want to read a pentest report critically. |
| OWASP Top 10 for LLM Applications | The 2025 list of risks for apps built on language models, from LLM01 Prompt Injection to LLM10 Unbounded Consumption. | Your app calls a model or runs an agent. |
| MDN: Security on the web | Mozilla's developer documentation, organized into attacks, defenses, threat modeling and authentication. | You want the browser's side explained: CSP, CORS, cookies, HTTPS. |
| CISA Known Exploited Vulnerabilities | The US agency's catalog of vulnerabilities known to be exploited. The catalog version of 13 November 2026 lists 1,781 entries. | You must decide which upgrade cannot wait. |
Two of these are often confused. The Top 10 names categories of risk and is meant to raise awareness. The ASVS is the document with requirements you can tick off, each with its own number, so a test report can say exactly which ones were verified.
For LLM features, our OWASP LLM Top 10 guide turns the ten risks into questions about your own app. For vulnerabilities in the frameworks that AI builders generate, the CVE watch keeps one page per CVE with the version that fixes it.
Free training with real labs
Reading about an attack and watching one work teach different things. Two free resources let you practise on targets built for it.
- [PortSwigger Web Security Academy](https://portswigger.net/web-security). Free online training from the makers of Burp Suite, with interactive labs and progress tracking. On 16 November 2026 its front page listed 30 labs on cross-site scripting, 16 on SQL injection, 5 on API testing and 7 on web LLM attacks.
- [OWASP Juice Shop](https://owasp.org/www-project-juice-shop/). An insecure web application built for security training, written in Node.js, Express and Angular. OWASP says it holds vulnerabilities from the entire OWASP Top Ten. You run it on your own machine and attack it there.
Checklists from the platforms themselves
The company that runs your backend knows its own defaults. Read its checklist before any third-party article, ours included.
| Platform | The page to read | What it covers | Our guide |
|---|---|---|---|
| Supabase | Production security checklist | The settings that, when missed, leave data exposed or accounts open to takeover, in Supabase's words. It includes a prompt that has an AI agent audit your project with read-only access. | Supabase security |
| Firebase | Firebase Security Rules | The rules that stand between your data and other users in Cloud Firestore, Realtime Database and Cloud Storage. | Firebase security rules |
| Next.js | How to think about data security in Next.js | Three ways to fetch data and what each assumes, with a data access layer recommended for new projects. | Next.js security |
| Stripe | Integration security guide | PCI compliance as a responsibility shared between Stripe and your business, and which integrations carry the least risk. | Stripe security |
A vendor checklist covers the vendor's side. Whether your own tables, rules and routes follow it is something only a test on your project can show.
Free and open-source tools, and the one question each answers
| Tool | The question it answers | Where it runs |
|---|---|---|
| MDN HTTP Observatory | Does my site follow good practice for security headers? Developed by Mozilla and launched in 2016. | Online, public addresses |
| Security headers checker | Which headers are missing or weak, and what is the exact line to add for Next.js, Vercel, Netlify, Cloudflare Pages or nginx? | Online, public addresses |
| keybrum-headers | The same header check from a terminal or a CI job, with the same grade. Open source under the MIT licence. | Your machine: localhost, staging, production |
| TruffleHog and Gitleaks | Is there a secret in my repository or its history? Our comparison says which to pick. | Your machine or CI |
OSV-Scanner and npm audit | Does a dependency I ship have a published vulnerability? | Your machine or CI |
| ZAP | What does a general web scanner find on my app? Free and open source. | Your machine |
| API key leak checker | Which keys in this JavaScript or env file are secret, and how do I rotate them? | Your browser, nothing sent |
| Supabase RLS checker | Which tables have row level security off, and which policies let anyone in? | Your browser, nothing sent |
| Next.js CVE checker | Do my Next.js and React versions have a known CVE, and which version fixes it? | Your browser, nothing sent |
Each of these answers one narrow question well. A header scan says nothing about your database. A secret scanner reads the repository you point it at, and a key can also sit in the bundle your live app serves. The pentest vs vulnerability scan guide explains the gap between listing weaknesses and proving one.
Guides for a specific job: data, search and LLM teams
Security looks different from each desk. A data engineer worries about a dashboard that exposes a warehouse. Someone who lives in Search Console worries about a hacked page dropping out of the index. A team that pays an LLM bill worries about a key that spends for a stranger. The guides below are written for those jobs.
For data teams (DataDef)
- Security checklist for data teams shipping web apps: twelve points for dashboards, APIs and analytics portals, from authentication to documenting the architecture.
- How data breaches start: lessons from exposed API endpoints: five incidents that needed no exploit, and six checks to run this week.
- OWASP Top 10 for data applications: the 2025 list in data terms, with a one-hour pass for a dashboard or an analytics API.
For search and launch (Porteur)
- Web security and SEO: why a hacked site loses rankings: what a compromise does to indexing and to Safe Browsing status, and the recovery steps in Search Console.
- Why Google downgrades hacked sites, and how to prevent it: how hacked content falls under Google's spam rules, how to notice early, and how long the way back takes.
- Security headers that improve both SEO and trust: what Google counts and what it does not, header by header.
- Run a security audit before you launch: what a launch exposes, when to test, and a pre-launch checklist.
For teams that run LLM features (Tokenwise)
- How to secure LLM API keys and endpoints in production: where keys leak, what a proxy layer must enforce, and a rotation routine.
- How to detect leaked LLM API keys in deployed web apps: the places a key ends up in what a browser downloads, and what to do once one is found.
- The hidden cost of a security breach on AI infrastructure: the bill beyond the tokens a thief burns, and the first hour after you suspect a breach.
- AI agent security: protect tool-calling endpoints: the three doors an agent with tools adds to an app, and how to check each.
On our side, the matching guides are the SaaS security checklist, what to do with a leaked API key and AI agent security.
What reading does not replace
Every resource above tells you what could be wrong. None of them looks at your app. A checklist says row level security should be on. Whether one of your tables answers a stranger is settled only by a request sent to the live app.
That request is what Keybrum sends. The free check reads what any visitor's browser already downloads from your app: the backend it talks to, the keys in the JavaScript and the tables it names, in about ten seconds and with no account. The paid attack run then tries more than 300 attacks, as a stranger and as accounts it creates, and reports each finding with the request that worked and a fix your coding agent reads through MCP. A one-time report is €49. Plans start at €29 a month excluding VAT and attack every deploy.
Questions
What is the best free resource to learn web application security?
PortSwigger's Web Security Academy is the most complete free course we know: written lessons with interactive labs, at no charge. Pair it with the OWASP Top 10:2025 for the map of risks. If you prefer one target to explore on your own machine, install OWASP Juice Shop.
Is the OWASP Top 10 a security checklist?
No. OWASP describes it as a standard awareness document: it names ten categories of risk and explains each. For requirements you can verify one by one, OWASP publishes the Application Security Verification Standard, at version 5.0.0 since 30 May 2025.
Which free tools can check my web app's security?
For response headers, the MDN HTTP Observatory or a security headers checker. For secrets in a repository, TruffleHog or Gitleaks. For known flaws in dependencies, OSV-Scanner or npm audit. For a general scan, ZAP. Each one answers a single question, so use several, and do not read a clean result from one as a verdict on the whole app.
Are these resources enough for an app built with AI tools?
They cover the theory and the configuration. Generated apps add a pattern of their own: the browser talks straight to Supabase or Firebase with a public key, so access control lives in policies and rules that are easy to leave open. Read the platform's checklist first, then test the live app as a stranger and as a second signed-up user.
How often should I come back to this list?
Reread the standards when they change: the current Top 10 is the 2025 edition, and the ASVS reached 5.0.0 the same year. Check the vulnerability catalogs when an advisory names a framework you ship. Tests are different: run them after every deploy, because a new table or route can reopen what the last test closed.
Where can I find security guides for data, SEO or LLM work?
This page lists three sets: DataDef's guides for data teams, Porteur's on what a compromise does to search rankings, and Tokenwise's on LLM keys and agent endpoints. All three are sister products of Keybrum.
Sources
- 1.OWASP Top 10:2025, OWASP
- 2.Application Security Verification Standard, OWASP
- 3.OWASP Cheat Sheet Series, OWASP
- 4.Web Security Testing Guide, OWASP
- 5.OWASP Top 10 for LLM Applications 2025, OWASP Gen AI Security Project
- 6.OWASP Juice Shop, OWASP
- 7.Web Security Academy, PortSwigger
- 8.Security on the web, MDN Web Docs
- 9.HTTP Observatory, MDN
- 10.Known Exploited Vulnerabilities Catalog, CISA
- 11.Production security checklist, Supabase Docs
- 12.Firebase Security Rules, Firebase Docs
- 13.How to think about data security in Next.js, Next.js Docs
- 14.Integration security guide, Stripe Docs
- 15.OSV-Scanner, Google
- 16.ZAP, ZAP