An authentication flow where one application obtains an access token to call another service while acting for a signed-in user. The second service receives a delegated token or assertion with the user’s identity and consent context, rather than the application’s own credentials, so access can be scoped to the user’s permissions and approved actions.
How On-Behalf-Of Authentication Works
An on-behalf-of flow is a delegated authentication pattern, not a simple sign-in. The application presents a token or assertion that proves the user already authenticated and approved the action, then receives a downstream token that is constrained to that user’s context. That design helps keep the caller from using its own broad credentials to reach a service.
The important distinction is that the downstream service should see a delegated identity, audience, and consent context, not a generic application credential. In practice, that makes the flow useful when an app must call another API with user-specific permissions while still preserving service-to-service trust boundaries.
For a standards-based view of the exchange itself, RFC 8693: OAuth 2.0 Token Exchange defines the core token exchange model used by on-behalf-of and related delegation patterns.
When teams need a broader conceptual map, OAuth Token Exchange and Delegation Guide explains how subject tokens, actor claims, and downscoping fit together in delegated access.
Delegation, Consent, and Authorization Boundaries
OBO flows matter because they preserve the user’s authorization context while letting one component act as a broker to another. That reduces the temptation to forward the original user token unchanged, and it gives the receiving service a clearer basis for authorization decisions tied to the user’s approved scope.
The model also introduces a boundary question: the application is not the user, even when it is acting for the user. Properly implemented, the flow should distinguish the acting application from the end user, which is why actor and subject semantics matter so much in delegation architectures.
That distinction is especially important in AI and automation scenarios where a system may act for a person but should not inherit unlimited reach. Agentic AI Identity Guide covers delegated authority, agent acting on behalf of a user, and token exchange as part of the identity model for agents.
Where implementers want a side-by-side comparison of patterns, AI Agent Service Accounts vs On-Behalf-Of Access shows when delegated user access is preferable to app-owned credentials and when it is not.
Security Properties and Common Failure Modes
The security value of OBO is that it can reduce over-privilege and make access easier to reason about, but only if token audience, token lifetime, and delegation semantics are enforced correctly. If the downstream service accepts a token meant for another audience, or if delegation is treated as an afterthought, the flow can collapse into bearer-token forwarding with all the usual replay and confused-deputy risks.
Another common failure mode is weak separation between the user session and the calling application’s own privileges. If the application can mint or reuse tokens without a strong delegation contract, a compromise in one layer can spill into other services that were never meant to accept that trust relationship.
For a practical comparison of these pitfalls, OAuth Token Exchange vs Token Passthrough explains why token passthrough weakens audience control and increases downstream replay risk.
Agent to SaaS API Access Patterns Guide is also useful when the same delegated-access question appears across SaaS integrations, where per-user tokens, app-only access, and delegation create very different exposure profiles.
Where OBO Fits in Modern Identity Architecture
OBO is most useful when the organisation wants to keep user intent visible across hops without handing every intermediary full access. It fits best in architectures where an application, API gateway, agent, or middle tier must act as a trusted intermediary while still preserving least-privilege constraints and traceability.
That makes it a natural companion to token exchange, consent-aware APIs, and delegated authorization design. It is not a universal answer, because some workflows are better served by application credentials or by explicit service accounts when no user context should flow downstream.
AI Agent Identity Management vs Traditional IAM helps place OBO inside the broader question of what carries over from workforce IAM and what must change for delegated, runtime access.
For the authentication layer underneath the delegation pattern, NIST SP 800-63 Digital Identity Guidelines provides the identity assurance and phishing-resistant authentication context that strengthens upstream user trust before delegation occurs.
Risk and Threat Considerations
OBO flows create security value only when the delegation boundary is precise. If tokens are forwarded too broadly, accepted by the wrong audience, or reused outside the intended consent context, the flow can turn into a high-value lateral movement path that preserves user authority while hiding the true actor.
Failure mechanism: Attackers or buggy integrations abuse delegation semantics, token audience confusion, or token replay to obtain downstream access that was never meant to outlive the original user action.
Impact: The result can be unauthorized API access, overbroad service reach, session replay, and difficult-to-trace compromise across services that trust delegated tokens.
Standards & Framework Alignment
This section maps relevant standards and security frameworks to the operational risks and controls described in this guidance.
NIST SP 800-63, NIST SP 800-53 Rev 5 and OWASP ASVS set the governance and control requirements practitioners need to meet.
| Framework | Control / Reference | Relevance |
|---|---|---|
| NIST SP 800-63 | Digital Identity Guidelines | Defines assurance and authentication context for delegated user sign-in |
| Recommendation — Apply strong user authentication and assurance before issuing delegated downstream access. | ||
| NIST SP 800-53 Rev 5 | IA-5 — Authenticator Management | Covers control and lifecycle of secrets and tokens used in delegated authentication |
| IA-8 — Identification and Authentication (Non-Organizational Users) | Supports delegated user identity handling when external users initiate the flow | |
| IA-9 — Service Identification and Authentication | Applies when services exchange tokens and authenticate to each other in OBO flows | |
| Recommendation — Manage token and credential lifecycle tightly for delegated access flows. Authenticate external users strongly before allowing on-behalf-of delegation. Authenticate downstream services and validate token audience before accepting delegated access. | ||
| OWASP ASVS | V10 — OAuth and OIDC | Directly addresses OAuth delegation, token exchange, and authorization grant handling |
| Recommendation — Verify OAuth delegation logic, audience checks, and token handling rules. | ||
Practitioner Guidance
Why practitioners should care: Treat OBO as an authorization boundary design, not just an OAuth implementation detail. The key decision is whether the downstream service should trust the user context, the calling application, or both, because that choice determines how much privilege can safely flow across the hop.
Common misunderstanding: Teams often assume any delegated token is automatically safer than an app credential. In reality, OBO is only safer when the token is audience-bound, short-lived, and restricted to the exact downstream action that the user approved.
Practitioner takeaway: Use OBO when user context must survive the hop, but pair it with strict audience validation, narrow consent scope, and clear separation between the caller’s authority and the user’s authority.
Related resources from NHI Mgmt Group
Deepen Your Knowledge
Free weekly newsletter
Subscribe to the NHI & AI Identity Journal
The latest on NHI and Agentic AI security – articles, research, breaches, news and events every week.
Bonus 33% off our NHI Course when you subscribe.
Reviewed and updated by the NHIMG editorial team on November 16, 2026.
NHI Mgmt Group — the #1 independent authority on Non-Human Identity, IAM, and Agentic AI security. nhimg.org