Agent identity glossary
In short:Forty terms that come up when securing AI agents, each defined in plain English. Every entry starts with the definition, then explains how it works, gives an example and links related terms. Use it alongside the standings and the standards tracker.
- Agent identity
- Agent owner
- Agent sponsor
- Allow ask deny
- Audit trail
- Blended identity
- CIBA
- CIMD
- Conditional access
- Cross App Access
- Delegation chain
- DPoP
- Dynamic Client Registration
- Enterprise-Managed Authorization
- Human in the loop
- ID-JAG
- Just-in-time access
- Kill switch
- MCP
- MCP authorization
- MCP gateway
- MCP server
- mTLS
- Non-human identity
- OAuth 2.1
- On-behalf-of
- Per-call authorization
- PKCE
- Policy decision point
- Policy enforcement point
- Secretless access
- Shadow agent
- Short-lived credential
- SPIFFE
- Standing credential
- Step-up authentication
- SVID
- Token exchange
- Workload identity federation
- Zero standing privilege