Per-request authorization means every action is checked individually instead of trusting a session once and assuming all later activity is safe. This matters for agents because the request that causes damage may be generated after the initial interaction has already been approved. It is a stronger fit for dynamic, tool-using identities.
What Per-Request Authorization Actually Changes
Per-request authorization moves the security decision from “was this session once approved?” to “is this specific action still allowed now?”. That shift matters because the risk surface is not the login event itself, but every subsequent call, tool invocation, or delegated action that can diverge from the user’s original intent.
It is especially relevant when permissions are dynamic, the target action is high impact, or an automated actor can chain many operations after the initial approval. In those cases, the safest design is to treat authorization as a live decision, not a one-time property of the session.
Why It Matters for Dynamic and Tool-Using Actors
Per-request authorization is a better fit for systems where authority can change mid-session, where one request can be harmless and the next destructive, or where an AI agent can call multiple tools under the same umbrella session. That is why action-scoped policy decisions and explicit approval boundaries are central to AI Agent Authorisation Guide and AI Agent Action Approvals and Spending Limits Guide.
This model also helps when the right answer depends on the exact request contents rather than on a broad user or workload role. A permission system that only checks the session at the start can miss a later attempt to access a new resource, change a limit, or trigger a side effect that was never part of the original approval.
How It Relates to Policy Models and Decision Points
Per-request authorization is usually implemented through fine-grained policy evaluation, externalized authorization, or request-level claims that describe the intended action in enough detail to decide it safely. The concept aligns naturally with Authorisation Models Guide because RBAC alone is often too coarse for action-level decisions, while ABAC, ReBAC, and policy-based approaches can express context, resource attributes, and relationship-aware checks.
It also connects to Externalised Authorisation Buyer's Guide, where the policy decision point and policy enforcement point are separated so each request can be evaluated independently. That separation is what makes it practical to inspect the action, context, and constraints before allowing execution.
Where It Shows Up in Real Systems
In API-heavy systems, per-request authorization is a control against broken object-level or function-level authorization, because each call must prove it can act on the specific object or function it is trying to reach. For cloud-native APIs, NIST SP 800-228 API Protection for Cloud-Native Systems Guide is a useful reference for request-level protection patterns, and the IETF’s RFC 6749: The OAuth 2.0 Authorization Framework remains foundational for understanding how delegated access is represented in practice.
For higher-assurance flows, request integrity and transaction specificity become even more important. OAuth Rich Authorization Requests (RAR) Guide and OAuth Pushed Authorization Requests and JAR Guide show how richer request descriptions and signed or pushed requests help ensure the policy engine evaluates the actual action, not an altered or vague request.
Risk and Threat Considerations
Per-request authorization reduces the chance that a safe-looking session is later reused for unsafe activity. The main danger is privilege drift inside an active session, where an attacker, over-privileged automation, or a compromised agent can keep issuing new actions that were never explicitly approved.
Failure mechanism: A session is treated as broadly trusted after the first approval, so later requests inherit authority they should not have, or the system fails to re-evaluate changed context, resource scope, or action class.
Impact: A single compromised session, token, or delegated agent can perform repeated unauthorized actions, expand access, or trigger destructive operations before the problem is detected.
Standards & Framework Alignment
This section maps relevant standards and security frameworks to the operational risks and controls described in this guidance.
OWASP API Security Top 10 addresses the attack and risk surface, while NIST SP 800-53 Rev 5 sets the governance and control requirements practitioners need to meet.
| Framework | Control / Reference | Relevance |
|---|---|---|
| NIST SP 800-53 Rev 5 | AC-3 — Access Enforcement | Per-request checks enforce each action against policy before access is granted. |
| IA-5 — Authenticator Management | Per-request authorization often depends on short-lived tokens and controlled credential use. | |
| AC-6 — Least Privilege | Action-by-action decisions prevent sessions from retaining broader authority than needed. | |
| Recommendation — Evaluate every request against current authorization policy before allowing the action. Limit token lifetime and validate credential use so each request remains bound to current authority. Constrain each request to the minimum privilege required for that specific action. | ||
| OWASP API Security Top 10 | API5 — Broken Function Level Authorization | Per-request authorization directly addresses function-level abuse by checking each action individually. |
| API1 — Broken Object Level Authorization | Request-scoped checks are needed to confirm each object access is allowed for the current request. | |
| Recommendation — Verify function-level authorization on every API call, not only at session start. Enforce object-level checks on every request that targets a resource instance. | ||
Practitioner Guidance
Why practitioners should care: Treat per-request authorization as the default for any workflow where the consequence of one request can differ materially from the next. The practical question is not whether a session is valid, but whether this exact request still deserves to proceed under current policy.
What to watch for: Watch for coarse session reuse, long-lived approvals, and policies that cannot express action-specific context. If a system cannot distinguish between harmless and harmful requests, it is relying on trust that is broader than the risk warrants.
Related resources from NHI Mgmt Group
- What is the difference between per-request authorization and authenticate once, trust indefinitely access?
- What is the difference between session-based access and per-request authorization?
- Why does per-request authorization improve incident response and compliance evidence?
- Why does per-request authorization reduce risk for workloads and agents?
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 5, 2026.
NHI Mgmt Group — the #1 independent authority on Non-Human Identity, IAM, and Agentic AI security. nhimg.org