Question 1

How is Cerberus different from a Web Application Firewall or API gateway?

Accepted Answer

Cerberus instruments inside the application instead of sitting in front of it. A WAF or API gateway has to infer intent from traffic shape; Cerberus sees the real request and response context, ties every event to a hashed user identity, and tracks behavior across services. The result is attribution, a single chain of activity per actor, rather than a queue of disconnected alerts.

Question 2

Does Cerberus send PII outside our environment?

Accepted Answer

No. PII is removed or hashed at the point of instrumentation, so raw values never leave the customer environment. This is the same property that makes cross-customer threat intelligence possible without identifiers crossing between environments, and it materially shortens vendor and privacy review cycles.

Question 3

What is cross-customer threat intelligence in API security?

Accepted Answer

Because every Cerberus customer instruments the same way, attack patterns observed at one institution are immediately used to harden every other customer. No raw identifiers cross between environments. Only the abstract attack pattern does. The practical effect is that emerging threats reach every customer's defenses in near-real time instead of waiting for vendor research and signature updates.

Question 4

How does Cerberus handle agentic AI and MCP traffic?

Accepted Answer

Cerberus handles agentic AI traffic and MCP tool calls natively, with the same in-app instrumentation model used for traditional API traffic. That means tool calls are tied to the same user-level attribution graph as the rest of the application, so prompt injection, exfiltration, and abuse of MCP servers exposed to untrusted content are detected as part of one chain of activity rather than as a separate, retrofitted bolt-on.

Question 5

What attack types does Cerberus detect?

Accepted Answer

Cerberus detects account takeover and credential abuse, business logic abuse, broken object-level authorization (BOLA), data scraping and exfiltration, bot-driven attacks, prompt injection through agentic AI tool calls, and abuse of MCP servers exposed to untrusted content. Detection runs against both a per-customer behavioral baseline and a cross-customer threat-intelligence layer.

Live in five minutes. Gateway or SDK.

Three steps to your first detection.

Install

Configure

Respond

Read the runtime. Run the harness.

Quiet by default. On in five minutes.