Topic starter
22/01/2026 2:28 pm
TL;DR: MCP use cases now span interactive, delegated, autonomous, and chained execution, and each one changes how identity, consent, and authorization have to work across AI agents and MCP servers, according to SGNL's discussion summary. The core issue is that agentic access is no longer a simple extension of human login flows, so governance has to follow the agent's authority and delegation model, not just the user session.
Editorial analysis by NHI Mgmt Group, based on content published by SGNL: “The four MCP use cases: Summary of a discussion at IIW”.
Key questions
Q: How should teams govern delegated MCP access when the user is no longer present?
A: Teams should govern delegated MCP access by binding it to the original task charter, not to a broad user session.
Q: Why do autonomous MCP agents need NHI-style identity controls?
A: Autonomous MCP agents need NHI-style controls because they operate as their own principals once deployed.
Q: What breaks when identity is not preserved across chained MCP requests?
A: Downstream authorisation breaks when the originating identity is lost, because the final server can only see the immediate caller.
Practitioner guidance
- Define a separate policy model for interactive agent sessions Treat user-present MCP sessions as consent-driven flows and require explicit approval for each sensitive step that the server surfaces during execution.
- Bound delegated tasks by charter and permitted subtasks Write delegation rules that describe what the agent may infer, what it may not infer, and which runtime requests must be denied when the user is absent.
- Register autonomous agents as first-class NHIs Assign durable ownership, scoped access and lifecycle records to autonomous agents instead of relying on the human who deployed them.
Bottom line: MCP use cases are no longer uniform, and IAM controls that assume one human principal and one session boundary are too narrow for interactive, delegated, autonomous and chained agent flows.
Explore further
View Full Forum → | NHI Foundation Course → | Our Services → | Read the full analysis →
This topic was modified 6 days ago by NHI Mgmt Group
06/10/2026 9:11 pm
Current IAM assumes a stable human principal behind the session, and MCP breaks that assumption. The interactive, delegated, autonomous and chained patterns show that access can no longer be evaluated only at login time. Once the actor can continue work after the user is gone, identity governance has to shift from session ownership to delegation state and principal continuity. Practitioners should stop treating agentic access as a thin extension of human SSO.
A few things that frame the scale:
- Gartner predicts that by 2028, 33% of enterprise software applications will include agentic AI, up from less than 1% in 2024, and that 15% of day-to-day work decisions will be made autonomously.
A question worth separating out:
Q: What is the difference between interactive and delegated MCP authorization?
A: Interactive MCP authorization depends on live user consent during the session, while delegated authorization depends on policy interpretation after the user has stepped away. The first can ask the human; the second must decide whether a new request still fits the original charter. That is a material governance difference, not just a timing difference.
👉 Read our full editorial: MCP identity patterns are outgrowing current IAM assumptions