// Security terms reference
SaaS security glossary
How to use this glossary
These are the security terms that come up most when an AI-built SaaS goes through review, due diligence or an incident. Each one is in a plain sentence, meant to answer "what is this?" without jargon. They are grouped by theme: access, injection, data, supply chain and method.
If a customer, a security questionnaire or a report used an acronym you did not recognize, it is probably here.
RET runs this test: a 25-hour Sprint for R$9,500, within 7 business days. See price and timeline for my case
Access and identity
| Term | What it is and why it matters |
|---|---|
| IDOR | Reaching another user data by swapping an identifier (the id in the URL or body) that the server does not verify as yours. The most common and cheapest flaw to exploit in SaaS. |
| BOLA | The same IDOR seen from the API side (Broken Object Level Authorization). It is the number 1 flaw in the OWASP API Top 10. |
| Broken Access Control | Any flaw where a user does something beyond their own permission. It is the number 1 category in the OWASP application Top 10. |
| Privilege escalation | Becoming admin or taking another role without the right, because the permission check is missing (or done only on the client). |
| Session hijacking | Taking over someone else session through a token that was stolen, predictable or not invalidated on logout. |
| MFA | A second factor beyond the password (app, hardware key, code). It blocks a login made with only a leaked credential, the credential-stuffing vector. |
Injection and untrusted input
| Term | What it is and why it matters |
|---|---|
| SQL Injection | Injecting a command into the database through an input that was not parameterized, reading or altering data it should not. |
| XSS | Injecting a script that runs in another user browser, stealing their session or acting in their name. |
| CSRF | Forcing the victim already logged-in browser to run an action (change e-mail, transfer) without their intent. |
| SSRF | Making the server fetch a URL chosen by the attacker, reaching the internal network or the cloud metadata (keys). |
| Command Injection | Running an operating-system command through an input that was not sanitized. |
| Prompt Injection | A malicious instruction embedded in content that diverts an AI agent from its goal. It is the number 1 flaw in the OWASP LLM Top 10. |
Data, secrets and isolation
| Term | What it is and why it matters |
|---|---|
| PII | Personally identifiable data (name, e-mail, document). Leaking PII is a privacy-law incident, not just a bug. |
| Secret exposure | An API key, token or password left in the client bundle, in a log or in the repository, becoming the attacker key. |
| Multi-tenant isolation | Making sure one customer data never reaches another. A failure here is the worst incident for a B2B SaaS. |
| Encryption in transit and at rest | TLS on the network and a cipher on storage, so data does not leak in transit or on disk. |
| Least privilege | Giving each user and service only the minimum access needed, so a compromised account reaches the least. |
Supply chain and execution
| Term | What it is and why it matters |
|---|---|
| Supply chain | Risk that enters through a third-party dependency, package or tool, not through the code you wrote. |
| Slopsquatting | Publishing a package with the name an AI hallucinates when suggesting code, hoping the dev installs it without checking. |
| Dependency confusion | Tricking the package manager into fetching a malicious public package instead of the internal one of the same name. |
| RCE | Remote code execution: the attacker runs their code on your server. The most severe flaw there is. |
| Insecure deserialization | Turning untrusted input into a system object, a classic path to reach RCE. |
Method, proof and disclosure
| Term | What it is and why it matters |
|---|---|
| Pentest | An authorized, manual penetration test that proves what breaks, with reproducible evidence. What RET does. |
| Risk Review | A paid human read of the risk flows, cheaper than a pentest, to find the obvious before scaling. |
| CVE | A public identifier for a known vulnerability (for example, CVE-2025-29927 in Next.js). |
| CVSS | A 0 to 10 score that measures a vulnerability severity, to prioritize the fix. |
| Responsible disclosure | Reporting a flaw to the owner and giving a fix window before making it public. How RET reports. |
| Zero-day | A flaw exploited before a vendor fix exists. It cannot be patched, only mitigated. |
Frequently asked questions
What is the difference between IDOR and BOLA?
They are the same flaw (reaching another object by swapping an id) seen from different angles: IDOR is the classic web-application term, BOLA is the name in the OWASP API Top 10. In practice, we treat them as the same risk.
Which flaw is the most severe?
RCE (remote code execution), because the attacker gets to run code on your server. Next come cross-customer leaks (isolation) and secret exposure, which grant broad access without needing RCE.
Does RET test all these terms?
RET looks for these classes of flaw in the manual pentest and Risk Review, focused on the flows that sustain revenue (login, billing, data, uploads, admin) and on isolation between customers. Every finding comes with reproducible evidence.
Who runs the review at RET?
Gabriel Lima Ferreira, founder and lead pentester, with an authorized scope and responsible disclosure.