Agents
An agent in ThunderID is a machine identity for autonomous software, such as an AI assistant, a background worker, or a scheduled job. It is a first-class identity, separate from any user, with its own identifier, owner, credentials, attributes, and lifecycle. Because an agent carries its own identity, every action it takes is attributable to it, and its access is granted, restricted, audited, and revoked independently of any user.
What ThunderID provides for an agent
- Create its identity
Define the attributes to collect and the onboarding steps administrators follow. Create an agent with its own owner and identity record.
Configure its credentials
Use a client secret or private key JWT so the agent can prove its identity when requesting tokens.
Limit its access
Assign roles and groups so the agent reaches only the resources its work needs.
Request the right token
Request a token for the agent's own work, for acting on behalf of a user, or for delegating a task to another agent.
Agent Token · On behalf of users · Agent-to-agent delegation
For the concepts behind agent identity, the subject, actor, and resource model, and why an agent needs its own identity class, see the AI Agents use cases. To connect a framework quickly, see Getting Started.
