Join our Newsletter — 33% off our NHI Course

Could delegated identity fix MCP access for agentic AI?

 

(@nhi-mgmt-group)
Member Moderator
Joined: 2 years ago
Posts: 20225
Topic starter  

TL;DR: Shared credentials and borrowed user identities are undermining auditability in MCP services, with Aryaka detailing how accountability breaks when one reusable secret stands in for delegated authority. Tool access needs narrower, protocol-aware enforcement so least privilege and recertification still work.

Editorial analysis by NHI Mgmt Group, based on content published by Aryaka: “Addressing the God Key Challenge in Agentic AI for MCP Servers — Effective Solutions Explained”.

Key questions

Q: Why do agentic workflows need delegated identity instead of static tokens?

A: Delegated identity lets the enforcement layer mint a short-lived token that matches the user, the task, and the target service.

Q: What is the difference between application access and protocol-aware authorization for MCP?

A: Application access answers whether the connection is allowed.

Q: What breaks when agents rely on shared credentials or borrowed user identities?

A: Auditability breaks first, followed by recertification and containment.

Practitioner guidance

  • Replace shared agent secrets with delegated identity Map every agent-to-tool path to a user-bound, short-lived credential model so downstream services can attribute actions to a human initiator and not just a service account.
  • Inspect MCP traffic at the enforcement layer Validate tool name, arguments and session context before forwarding requests so policy can distinguish low-risk actions from high-risk tool invocations.
  • Remove broad OAuth scopes from agent runtimes Scope credentials to the target MCP service and the specific operation, then deny reuse of generic tokens that can reach unrelated tools or repositories.

Bottom line: Agentic AI creates an identity governance problem when shared credentials stand in for delegated authority and tool-level policy.

Glossary terms: Delegated Identity, Protocol-aware Authorization, Identity propagation, Blast Radius

Explore further

View Full Forum →  |  NHI Foundation Course →  |  Our Services →  |  Read the full analysis →


This topic was modified 1 week ago by NHI Mgmt Group

   
Quote
(@mr-nhi)
Member Moderator
Joined: 6 months ago
Posts: 20715
 

Delegated identity is now the baseline control, not an optional enhancement: agentic workflows break the old assumption that one stable credential can safely represent both a user and a tool-using runtime. Once an agent acts across multiple tools, identity has to be reconstituted at the enforcement layer instead of being inherited from a shared secret. The practitioner implication is that auditability now depends on delegated identity design, not on log review after the fact.

A few things that frame the scale:

  • 67% of organisations still rely heavily on static credentials despite the risks they pose to agentic AI deployments, according to the 2026 Infrastructure Identity Survey.
  • 24,008 unique secrets were exposed in MCP configuration files in 2025 alone, the protocol's first year of widespread adoption, according to the State of Secrets Sprawl 2026.

A question worth separating out:

Q: Who should own policy when agents call internal tools at scale?

A: Policy should sit at the enforcement layer that can validate the user, inspect the tool request, and issue a narrowed credential before forwarding traffic. Putting the logic in every MCP service creates inconsistent controls and pushes enterprise governance into places that were never built to hold it.

👉 Read our full editorial: MCP-aware Zero Trust is now the control point for agent access


This post was modified 1 week ago by NHI Mgmt Group

   
ReplyQuote
Share:

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.