Knowledge base
The AI and cloud security library
Practical, vendor-neutral explainers on AI, API, web, container and cloud security. All 207 are free to read, with no login. New here? Start with path 01, then work across.
Filter by topic (137)
-
Data and Model Poisoning
Most attacks on an application target it at runtime. Poisoning targets it earlier, during learning.
-
DLP That Works Without Blocking the Business
Data Loss Prevention has a reputation problem, and it earned it. Too many deployments arrive as a wall of blocked emails, quarantined uploads.
-
Integrating Security Scanning Into CI/CD Gates
Automating a security scan is easy. Deciding what it should be allowed to stop is the hard part.
-
Kubernetes Network Policies and Zero-Trust East-West Traffic
Ask most teams how their pods are segmented and the answer is: they aren't. Perimeter firewalls see none of this east-west traffic.
-
Pentest vs Red Team vs Bug Bounty: Choosing the Right Offense
Penetration tests, red teams and bug bounties all use authorized attackers but answer different questions, cost differently and suit different maturity levels.
-
Privileged Access Management, Without the Vault Theater
Privileged access is where a breach stops being an incident and becomes a catastrophe.
-
Security Misconfiguration: Insecure by Default
Security Misconfiguration is the category that catches everyone eventually, because it does not require a mistake in your code at all, only a default.
-
The Agentic Loop and Where It Breaks
A single model call is a function: text in, text out. An agent is a loop wrapped around that function, it perceives a situation, plans a step, acts.
-
Unrestricted Resource Consumption: The Bill and the Blackout
Every API request costs CPU, memory, bandwidth and sometimes money. Unrestricted Resource Consumption is the risk that an API sets no limit on that cost.
-
BFLA: When Users Reach Admin Functions
If Broken Object Level Authorization is about which records a user can reach, Broken Function Level Authorization is about which actions they can perform.
-
EDR and XDR: Endpoint Detection, Explained
The move from antivirus to EDR is the move from nouns to verbs. Antivirus inspected files and asked whether each matched a known-bad signature.
-
Identity-First Security: The New Perimeter Is Who You Are
Once you accept that network location grants no trust, a question follows immediately: what does?
-
Over-Permissioned Roles and Privilege Escalation Paths
A cloud attacker who lands on a low-privilege role does not go looking for an admin password.
-
Privileged Containers and Container Escapes
Containers isolate with namespaces and cgroups, not with the hardware-enforced boundary a virtual machine gets.
-
Prompt Injection, Explained
Prompt injection is the SQL injection of the AI era: a class of attack where untrusted text ends up being treated as trusted instructions.
-
VPN vs ZTNA: Beyond the Network Perimeter
The VPN was designed for a world that no longer exists, a trusted office network, a few remote workers dialling in, a hard perimeter around it all.
-
Vulnerable and Outdated Components: The Risk You Inherited
OWASP's Vulnerable and Outdated Components risk covers inherited code with a publicly known vulnerability. It rose from ninth to sixth place in the 2021 Top 10.
-
Agent Memory and Persistence Risks
A stateless model forgets you the moment the conversation ends. That is safe and useless in equal measure, it cannot remember your preferences.
-
Encryption and Key Management, Without the Hand-Waving
Asking whether the data is encrypted is the wrong question, because the answer is almost always yes and it tells you almost nothing.
-
Identification and Authentication Failures: Proving Who You Are
Once ranked second as Broken Authentication, this category was renamed Identification and Authentication Failures and moved to seventh in the 2021 OWASP Top 10.
-
Improper Output Handling: When Model Output Becomes an Exploit
Every developer learns not to trust user input. LLM05 asks a harder question: do you trust the model's output?
-
Infrastructure-as-Code Security: The Misconfiguration Factory
Infrastructure as code was one of the best things to happen to operations, and it quietly changed the shape of cloud security.
-
Joiners, Movers, Leavers: The Identity Lifecycle Gap
Every identity has a life: it is created when someone joins, changed as they move between roles, and, in theory, removed when they leave.
-
Why an Orchestrator, When You Already Have MCP
A common misread of the modern agent stack is that MCP is the agent. MCP is a protocol for advertising and invoking tools.