Agent authentication
Agent authentication is how a system knows which AI agent is calling it. Each agent signs in as itself, with its own credential, instead of borrowing a person’s login or a shared key.
In a run
Before the first step, Ares issues the agent a credential for this job only. Every call the agent makes carries it, and the record names the agent on every line.
Our word for it
Credentials for agents. Each agent signs in as itself.
Why a shared key is not enough
When several agents share one key or one person’s login, the record cannot say which of them acted. A refusal cannot be traced to a job, and revoking access for one agent stops all of them. With its own credential, each agent can be checked, refused and revoked on its own.
Where the identity comes from
The identity comes from your directory. See identity provider for which directories work today. Agents you already run are enrolled without a rewrite: see bring the agents you have.
What the record shows
Each line names the agent, the person or event it acted for, and the credential it used. A refused call names the agent that made it, so every refusal points to one agent and one job.
Where it stops
- Agent authentication says who is calling. It does not say what the call may do. That is the job of the access policy.
- Agents you have not enrolled in Ares are not covered.
Run the sandbox to see a credential issued for one job, and a call refused under it.
Related terms
- Agent identityAgent identity is what makes an AI agent one named party that every system can recognise, separate from the people and other agents around it.
- Scoped credentialA scoped credential covers a named set of actions for a limited time, instead of everything an account can do.
Tell us about one workflow.
We will say whether an agent should do it, and what the first step costs.
A senior member of the team replies within two business days.