Glossary

This glossary provides definitions for key terms and concepts used throughout the Securix.app documentation.

A

Agent Access Security Broker (AASB) : The core architectural pattern of Securix.app. It acts as a secure intermediary (a "Trust Buffer") that brokers access between AI agents and third-party data sources, replacing traditional OAuth flows and preventing direct exposure of sensitive tokens to agents.

API Key : A unique identifier used to authenticate requests to the Securix.app API. Client-side API keys typically have restricted permissions and are exposed to the browser, while server-side API keys are highly sensitive and must be kept confidential.

Authorization : The process of determining whether an authenticated user or AI agent has permission to perform a specific action on a particular resource. Securix.app's policy engine handles this based on defined policies.

C

Client SDK (@securix/client) : The software development kit designed for client-side applications (web, mobile) to interact with Securix.app services for UI-driven security features and initiating secure connections.

Cloud Connector : An integration within the Multi-Cloud Platform (MCP) that allows Securix.app to securely communicate with and manage resources within a specific cloud provider (e.g., AWS, Azure, GCP).

Core SDK (@securix/core) : The software development kit designed for server-side and backend applications. It provides comprehensive features for policy enforcement, multi-cloud management, and acting as the "Trust Buffer" for AI agents.

D

Data Redaction : The process of removing or masking sensitive information from data before it is presented to a user or AI agent, based on defined security policies.

M

Model Context Protocol (MCP) Gateway : A feature of Securix.app that enables AI agents to access data and services through a secure, policy-driven interface. It ensures zero-token exposure for agents and allows for real-time control over data access.

Multi-Cloud Platform (MCP) : Securix.app's unified control plane for managing security policies, identity and access control, and threat detection across diverse public and private cloud environments.

P

Policy : A set of rules defined within SecuriX that specifies who (or what, like an AI agent) can perform which actions on what resources under what conditions.

Policy Enforcement Point (PEP) : The SecuriX Gateway component that intercepts agent tool calls and evaluates them against defined policies before allowing them to reach the provider.

Policy Abstraction Layer : A component of MCP that allows you to define high-level, cloud-agnostic security policies, which are then translated into native security controls for each connected cloud.

S

Self-healing Auth : A unique SecuriX capability where the Gateway returns structured metadata (llmError) to an AI agent, allowing the agent to explain connection or policy issues to the user and provide direct links for resolution.

T

Throttling : The process of limiting the rate at which a user or AI agent can access a resource or perform an action, based on defined security policies, to prevent abuse or overload.

Trust Buffer : A conceptual and architectural component of Securix.app (implemented by the AASB and MCP Gateway) that acts as a secure intermediary, preventing direct exposure of sensitive credentials (like OAuth tokens) to AI agents while still allowing them to access necessary data and services.

Unified Control Plane : A centralized interface provided by MCP for managing security policies and operations across all integrated cloud environments.