Skip to main content
Oathera logo Oathera
Use casesFive stories: no identity, a new identity, impersonation, cloning, and a call outside the boundary. PlatformOathera Agent Enrollment, the Oathera Identity Service and Oathera Agent Enforcement. FeaturesWhat Oathera does for you, grouped by purpose. IntegrationsYour identity provider, your MCP servers, your cloud. SecurityHow Oathera fails closed, and what it doesn't do.
Developer sectionComponents, the identity token, standards and settings. Data sheetsThe buyer and developer data sheets, as PDFs. LearnGuides on giving AI agents an identity of their own. GlossaryPrincipal, sponsor, anchor key, key digest, identity token. SimulatorWatch a stolen token, a replayed proof and a tampered call get refused.
About usThe team and the mission behind Oathera. BlogPosts from the Oathera team. CareersOpen roles at Oathera. ContactTalk to the founders, or become a design partner. LinkedIn ↗Oathera on LinkedIn.
Privacy PolicyHow we handle data. Terms of ServiceTerms for using Oathera. Data Processing AgreementOur DPA for customers. Sub-processorsThird parties we rely on.
Become a design partner

Learn

Know Your Agent in production

Straight answers to the questions security and platform teams ask when real AI agents start touching real systems: how to give each agent its own identity, how to scope and revoke access, how to prove which agent did what, and where existing tools stop short. Each guide is self-contained and practical.

Become a design partner
Illustration of an open book with a rocket, flask, light bulb and other learning icons

Topics

  1. Identity
  2. Governance & accountability
  3. Access & enforcement
  4. Standards & interop
  5. Terms & definitions
  6. Threats, controls & standards
  7. Oathera compared
  8. Try it

Identity

  • Give each AI agent its own identity instead of a shared API key

    Why a shared key can't tell your agents apart in logs, and how to issue a distinct, verifiable identity per agent.

  • Make a named person accountable for an AI agent

    Tie every agent to the person who approved it. Record the sponsor who answers for it.

  • Identity tokens vs standing API keys

    The real security tradeoff for AI agents, and whether the operational overhead is worth it.

Governance & accountability

  • What agentic identity governance looks like

    Who is accountable when an AI agent takes a destructive action, and the controls that make the answer provable.

  • Audit every action an AI agent takes

    Prove the identity behind each request instead of just logging a token that anyone could have copied.

Access & enforcement

  • Enforce least privilege for an AI agent calling your systems

    Scope an agent to the smallest set of operations it needs, and enforce it on every request.

  • Revoke an AI agent's access immediately

    A playbook for cutting off an agent the moment it starts behaving unexpectedly.

  • Operational boundaries, enforced at runtime

    What an operational boundary is, and why enforcing it at runtime beats trusting a config file.

Standards & interop

  • Oathera's own rules: observe first, then enforce

    What a rule decides, why each one watches before it enforces, and which rule wins.

  • SPIFFE/SPIRE and the agentic identity gap

    Workload identity covers services; here's what's missing for an agent acting on a human's behalf, and how teams fill it.

Terms & definitions

  • What is agentic identity? Agent identity, defined

    Agentic identity (also called AI agent identity, agent identity or autonomous agent identity) is an identity an AI agent holds in its own right: bound to where it…

  • What is Know Your Agent (KYA)? KYA verification explained

    Know Your Agent (KYA) is the practice of verifying which AI agent is acting, which person answers for it and whether it may act, before it acts. KYA verification is…

  • Non-human identity (NHI) governance for AI agents

    Non-human identity (NHI), non-person entity (NPE) and machine identity cover service accounts, keys and certificates. AI agents add a named accountable person and…

  • Agentic IAM: identity and access management for AI agents

    Agentic IAM is identity and access management built for AI agents: a per-agent identity, a named accountable person, runtime authorization on every call, and…

  • Proof of possession for AI agents: anchor keys, session keys and operation proofs

    Proof of possession (PoP) for AI agents means a token is useless without the key it names. How a non-exportable anchor key, a rotating in-memory session key and a…

  • MCP security and authorization for AI agents

    Model Context Protocol (MCP) servers expose tools to AI agents. How to secure them: authenticate each agent, authorize each tool call, and verify per-call proofs at…

Threats, controls & standards

  • Confused deputy and prompt-injection privilege escalation in AI agents

    A prompt-injected AI agent can act as a confused deputy, using its own access for an attacker. Why detection alone fails and how per-agent identity, narrow boundaries…

  • Stop token theft and replay attacks on AI agents, and retire shared API keys

    Stolen tokens and shared API keys let anyone act as an agent. How per-agent keys, rotating session keys and single-use operation proofs stop token theft and replay,…

  • Shadow AI agents: discover, enroll or reject

    Shadow AI agents are agents running without approval or identity. How to find agents calling your tools with no identity, and enroll or reject each one.

  • Contain blast radius: over-privileged service accounts and just-in-time agent access

    Over-privileged service accounts make every agent compromise worse. How to contain blast radius with per-agent identity, narrow operational boundaries, audience-bound…

  • Multi-hop delegation and agent-to-agent authentication

    When one AI agent hands work to another, who authorized what? How signed task context, narrowing-only delegation, single-use approvals and agent-to-agent identity…

  • AI agent audit trails and EU AI Act accountability

    What an AI agent audit trail should record, how deterministic provenance supports accountability and record-keeping, and what the EU AI Act does and does not ask of…

  • Standards map for AI agent identity: what Oathera uses, and what it does not

    Which standards apply to AI agent identity: OIDC with PKCE, JWT and JWS, RFC 7800, RFC 7638, RFC 9421, W3C verifiable credentials, SPIFFE, OWASP NHI and agentic…

Oathera compared

  • Oathera vs Okta, Microsoft Entra and Auth0 for AI agents

    Okta, Microsoft Entra and Auth0 are built around human sign-in and OAuth clients. Oathera adds what AI agents need: an anchor key on the agent's host, a single-use…

  • Oathera vs SailPoint, CyberArk and Saviynt for AI agents

    SailPoint, CyberArk and Saviynt centre on identity governance reviews and credential vaulting. Oathera checks AI agents at runtime: per-call proof and decisions, with…

  • Oathera vs AWS IAM, Google Cloud IAM and Azure managed identities for AI agents

    AWS IAM, Google Cloud IAM and Azure managed identities give workloads identities tied to cloud infrastructure. Oathera gives an AI agent one identity bound to its…

  • Oathera vs Palo Alto Networks and IBM security suites for AI agents

    Broad security suites from vendors such as Palo Alto Networks and IBM watch networks, endpoints and events. Oathera is a focused identity layer for AI agents:…

Try it

  • Prompt Heist: talk your way past three AI agents

    A free browser game. Use direct injection, indirect injection and approval replay on three fictional agents, then see what stops each attack.

  • Data Vault Breach: make three AI assistants give up their secrets

    The second chapter in the series. Use prompt leakage, claimed authority and output filter bypass, then see what stops each.

  • Inbox Infiltrator: plant text that three AI assistants obey

    The third chapter in the series. Use indirect injection, the confused deputy and hidden page text, then see what stops each.

  • Tool Chain Saboteur: break the guards around an agent's tools

    The fourth chapter in the series. Hijack a tool, smuggle an argument, chain safe tools into a leak, then beat a boss act.

  • Multi-Agent Conspiracy: attack the trust inside a team of agents

    The fifth chapter in the series. Inject across agents, forge a message, reuse an approval, then chain them in a boss act.

  • Identity Shadow: what still works against a hardened agent setup

    The finale of the series. Attack a hardened setup: in-scope misuse, a path around enforcement and identity confusion, then a boss.

  • What still works against a hardened AI agent

    What is left to attack when a gateway checks every call, and what closes each gap.

  • How multi-agent AI systems get compromised

    How cross-agent injection, forged messages and weak approvals work, and the fix for each.

  • How AI agent tool calls get abused

    How tool hijacking, argument smuggling and tool chaining work, and the fix for each.

  • How AI agents get hijacked by what they read

    Why emails, tickets and web pages can steer an agent, and the fix for each.

  • How AI assistants leak what they were told to hide

    Why a rule saying "keep this secret" fails against a persuasive request, and the fix for each failure.

  • Why AI agents fall for prompt injection, and what stops it

    The three attacks behind the game, why each one works on an agent with no verified identity behind its actions, and the fix for each.

← Back to Oathera
Oathera logo Oathera

Every AI agent, sworn in. Oathera gives every AI agent an identity bound to the machine it runs on, with a named person who answers for it, and checked every time it acts.

Product

  • Use cases
  • Platform
  • Features
  • Integrations
  • Security

Developers

  • Developer section
  • Data sheets
  • Learn
  • Glossary
  • Simulator

Company

  • About us
  • Blog
  • Careers
  • Contact
  • GitHub

Legal

  • Privacy Policy
  • Terms of Service
  • Data Processing Agreement
  • Sub-processors
© 2026 Oathera Inc. · All rights reserved
Know Your Agent (KYA)

Cookie preferences

We use cookies to run this site and, with your consent, to understand usage and improve Oathera. Strictly necessary cookies are always on; you can choose whether to allow analytics and marketing cookies below.

Read the privacy policy

  • Strictly necessaryAlways on

    Required for the site to work, security, load balancing, and remembering your cookie choices. These cannot be switched off.

  • Help us measure traffic and see how the site is used, so we can improve it. No personal profiles are built.

  • Used to make messages about Oathera more relevant across other sites. Off unless you turn it on.