Introducing the Simple Agent Management Protocol

AI agents are becoming operational systems. They discover resources, invoke tools, exchange tasks, retain memory and increasingly act with varying degrees of autonomy.

But as agent environments grow, a fundamental operational question remains:

How do we consistently discover, monitor, configure and govern heterogeneous AI agents across different frameworks, vendors and deployment environments?

The Simple Agent Management Protocol (SAMP) is an experimental proposal for addressing that problem.

An operational management plane for AI agents

SAMP defines a lightweight, vendor-neutral protocol between a management system and managed AI agents.

Its purpose is not to define how agents reason, collaborate or perform tasks. Its purpose is to provide the operational mechanisms required to understand and control an agent environment.

The initial protocol defines capabilities for:

• agent discovery and enrollment;
• health, state and profile inspection;
• operational event reporting;
• event-stream subscriptions;
• controlled configuration;
• lifecycle and trust-state management;
• and optional policy-gated execution.

Complementary, not competitive

The emerging agent-protocol ecosystem contains several distinct relationships.

Agent-to-agent protocols address communication, collaboration, delegation and task exchange. MCP addresses access to tools, resources and context. SAMP addresses another layer: manager-to-agent operational management and control.

These functions are complementary. A production agent environment may require all of them.

SAMP is also influenced by established operational-management approaches such as SNMP and OpAMP, while introducing concepts required for AI agents, including dynamic capability profiles, autonomy classes, enrollment, explicit trust states and restrictive execution policies.

Experimental by design

SAMP version 0.1 is intentionally Experimental.

It provides a concrete protocol baseline that can be reviewed, challenged, implemented and tested. The objective of the initial version is not to declare the problem solved, but to establish enough structure for meaningful technical discussion and interoperability work.

The first Internet-Draft was published on 17 August 2026:

https://datatracker.ietf.org/doc/draft-efstathiou-samp-agent-management/

This is an individual Internet-Draft. It is not endorsed by the IETF and currently has no formal standing in the IETF standards process.

What comes next

The next stage of the project will focus on:

• technical review of the protocol model and scope;
• validation of operational and security requirements;
• reference implementations and framework adapters;
• controlled proof-of-concept deployments;
• and independent interoperability testing.

Engineers, protocol designers, agent-framework developers, researchers and operators are invited to examine the draft and identify missing requirements, unnecessary complexity, security weaknesses and areas of overlap with existing work.

Technical criticism is not merely welcome. It is necessary.

Contact: [email protected]