Glossary

Definitions for terms, acronyms, and platform names used throughout the Agent Identity Security documentation.

Terms A–Z

Use the Category column to filter or sort by topic area.

Term Category Definition
Agent discovery Agent Identity Security platform The process by which Agent Identity Security automatically detects AI agent sessions running on a machine or inside a platform. Discovery is triggered when a Guardian or SDK integration is active and a new session starts.
Agent Identity Security Agent Identity Security platform The AppViewX platform that provides identity, governance, and security controls for AI agents across development and production environments. Agent Identity Security discovers agents, applies policies, and maintains an audit trail of every session.
API (Application Programming Interface) General technology A defined contract that allows software systems to communicate with each other. Agent Identity Security exposes REST APIs for integration with external platforms and provides API keys to authenticate those integrations.
API key Agent Identity Security platform A credential you generate in Agent Identity Security to authenticate Guardian installations and SDK integrations. Each key is assigned one or more scopes that limit what the key can do. Copy the key immediately after generation — Agent Identity Security does not store it and cannot display it again.
Audit log Agent Identity Security platform A tamper-evident record of every action taken by an AI agent session: tool calls, file accesses, policy decisions, and session lifecycle events. Audit logs are visible in Event Logs in the Agent Identity Security console.
AWS EKS (Amazon Elastic Kubernetes Service) Cloud and infrastructure A managed Kubernetes service from Amazon Web Services. Agent Identity Security can be deployed on EKS for on-premises or private-cloud installations, and AI agents running in EKS workloads can be connected to Agent Identity Security through the SDK or non-coding agent integrations.
Claude Code Agent platform An AI-powered terminal coding assistant developed by Anthropic. Agent Identity Security connects to Claude Code sessions using the Claude Code Guardian.
Claude Desktop Agent platform The desktop application version of Anthropic’s Claude AI assistant. Agent Identity Security connects to Claude Desktop sessions using the Claude Desktop Guardian.
CLI (Command-Line Interface) General technology A text-based interface for interacting with software by typing commands in a terminal or shell. Several Agent Identity Security-supported agents (such as Claude Code and GitHub Copilot CLI) operate primarily as CLI tools.
CrewAI Agent platform An open-source Python framework for building and running multi-agent AI systems. Agent Identity Security connects to CrewAI projects through the CrewAI SDK, which is added directly to a Python project rather than installed at the machine level.
Cursor Agent platform An AI-powered code editor developed by Anysphere. Agent Identity Security connects to Cursor sessions using the Cursor Guardian.
Event Logs Agent Identity Security platform The Agent Identity Security console page that displays real-time and historical audit events from all connected agents. Use Event Logs to investigate activity, verify policy enforcement, and export records for compliance reporting.
GitHub Copilot CLI Agent platform A command-line interface extension for GitHub Copilot that brings AI assistance to the terminal. Agent Identity Security connects to GitHub Copilot CLI sessions using the GitHub Copilot CLI Guardian.
gitleaks Cloud and infrastructure An open-source tool that scans code repositories and text for secrets such as API keys, passwords, and tokens. The Claude Code Guardian downloads and uses gitleaks during installation to strengthen secret-detection capabilities at the session level.
Guardian Agent Identity Security platform A lightweight background process installed on a developer workstation. The Guardian intercepts AI agent sessions, reports events to Agent Identity Security, and enforces the policies assigned to that machine. Each supported coding agent has its own Guardian package (for example, Claude Code Guardian).
HITL (Human-in-the-Loop) Agent Identity Security platform An approval workflow in which a human reviewer must authorize certain agent actions before they are executed. In Agent Identity Security, HITL approval is configured through a policy and requires an ITSM or notification integration.
IDE (Integrated Development Environment) General technology A software application that combines a code editor, debugger, and build tools in a single interface. Cursor is an example of an AI-enhanced IDE supported by Agent Identity Security.
ITSM (IT Service Management) General technology A set of policies and practices for managing IT services. In Agent Identity Security, ITSM integrations (such as Jira Service Management or ServiceNow) are used to route HITL approval requests and incident tickets.
jq Cloud and infrastructure A lightweight command-line JSON processor. The Guardian installer uses jq to parse configuration responses from the Agent Identity Security API during the setup process. It is downloaded automatically if not already present on the machine.
MCP (Model Context Protocol) Agent Identity Security platform An open standard that allows AI models to call external tools and data sources in a structured, auditable way. Agent Identity Security can monitor and restrict MCP tool calls through the MCP Gateway.
MCP Gateway Agent Identity Security platform An Agent Identity Security component that acts as a secure proxy between an AI agent and its MCP tool servers. All MCP traffic passes through the Gateway, where Agent Identity Security can inspect, allow, or block individual tool calls and apply data sanitization rules such as PII and secret scanning.
MDM (Mobile Device Management) General technology A category of software used to deploy, configure, and manage devices across an organization. Guardian packages can be distributed to developer workstations at scale using MDM tools such as Jamf Pro or Microsoft Intune.
On-Prem (On-Premises) General technology A deployment model in which software runs on servers that your organization owns and manages. Agent Identity Security supports on-premises deployment as an alternative to SaaS. During Guardian installation, select On-Prem and enter the full URL of your Agent Identity Security instance.
PII (Personally Identifiable Information) General technology Any data that can identify a specific individual, such as names, email addresses, phone numbers, or national ID numbers. The Agent Identity Security MCP Gateway can scan agent inputs and outputs for PII and block or redact it before it reaches an external tool server.
Policy Agent Identity Security platform A named set of rules in Agent Identity Security that controls what an AI agent is permitted to do. Policies can restrict MCP tool calls, file access, network destinations, and more. Policies are created in Secure → Policies and applied to agents through the Registry.
RBAC (Role-Based Access Control) General technology A security model in which access rights are assigned to roles rather than individuals. Agent Identity Security uses RBAC to control which users can view agents, manage policies, read audit logs, or administer API keys.
Registry Agent Identity Security platform The Agent Identity Security console section that lists all discovered agents, their configurations, assigned policies, and execution history. Use the Registry to review the inventory of agents connected to your Agent Identity Security tenant.
REST (Representational State Transfer) General technology A widely adopted architectural style for designing HTTP-based APIs. Agent Identity Security exposes REST APIs for agent management, policy retrieval, and audit log access, used by non-coding agent integrations such as Amazon Bedrock Agents and OpenAI.
SaaS (Software as a Service) General technology A cloud-hosted software delivery model in which the vendor manages infrastructure and updates. The default Agent Identity Security deployment is SaaS, hosted at https://<tenant>.stage.appvx.ai.
Scope (API key scope) Agent Identity Security platform A permission attached to an API key that limits what operations the key can perform. Guardian and SDK integrations require three scopes: audit_logs:write, agent_discovery:write, and policy:read.
SDK (Software Development Kit) General technology A packaged set of libraries, tools, and documentation that developers use to integrate their application with a platform. Agent Identity Security provides an SDK for Python-based agent frameworks such as CrewAI, enabling integration without a machine-level Guardian install.
SDK hooks General technology Callback points within the Agent Identity Security SDK that fire at key moments in an agent’s lifecycle (for example, session start, tool call, session end). SDK hooks capture event data and forward it to Agent Identity Security for auditing and policy enforcement.
SIEM (Security Information and Event Management) General technology A platform that aggregates and correlates security event data from across an organization’s infrastructure. Agent Identity Security audit logs and event streams can be forwarded to a SIEM (such as Splunk or Microsoft Sentinel) for centralized threat detection and compliance reporting.
SSL / TLS (Secure Sockets Layer / Transport Layer Security) General technology Cryptographic protocols that secure network communication. During Guardian installation, the SSL verification step configures how the Guardian validates the Agent Identity Security server’s certificate. For SaaS, select System CA. For on-premises, point to your organization’s internal certificate authority file.
Tenant ID Agent Identity Security platform The unique identifier for your Agent Identity Security organization on the SaaS platform. For most customers it is the company subdomain in the Agent Identity Security URL (for example, acme from https://acme.stage.appvx.ai). Entered during Guardian installation to point the agent to the correct Agent Identity Security instance.
WAF (Web Application Firewall) Cloud and infrastructure A security control that monitors and filters HTTP traffic to and from a web application. In Agent Identity Security deployments, a WAF is commonly placed in front of the Agent Identity Security server to protect against common web-layer attacks such as SQL injection and cross-site scripting.
Zero Trust General technology A security model based on the principle of never implicitly trusting any user, device, or workload, regardless of network location. Agent Identity Security applies Zero Trust principles to AI agents by requiring every session to be discovered, authenticated, and governed before it can access tools or data.