liveBook · Manning

Society of agents

As explained in Authorization in Action

A society of agents is a cooperative environment in which independent software systems interact under shared rules, protocols, and governance mechanisms.

A society of agents forms when independent software systems cooperate across organizational boundaries under shared rules, protocols, and governance. Unlike a centrally controlled environment, each organization retains its own policies and operational controls, so reliable cooperation depends on explicit commitments and verifiable limits. section 18.7

Cross-domain cooperation combines several controls:

  • Policies restrict what each agent may do within its own domain. section 18.7 section 18.8
  • Promises state how a receiving agent intends to behave, including actions, constraints, audit behavior, and policy framework. section 18.7
  • Credentials transport delegated authority across domains, while the receiving organization’s local policy still decides whether an action is permitted. section 18.8
  • Reputation and shared audit outcomes help a requesting agent assess whether a partner’s promises are reliable. section 18.7

Contractor reimbursements

An ACME reimbursement agent works with a financial partner’s payment agent. ACME issues a credential allowing contractor reimbursement payments up to $250 in USD, for specified expense categories, with receipts and an expiration date. The partner evaluates each request under its own policy, returns a signed payment receipt, and both domains retain the outcome for later governance and reputation decisions. section 18.7

How Authorization in Action covers it

How deep it goes
Explains policies, promises, credentials, reputation, and audit outcomes through a cross-domain reimbursement scenario, without detailing implementation internals.
Examples
Uses contractor reimbursements, payment agents, constrained credentials, policy rules, signed receipts, and cross-domain audit logs.
What it assumes
Assumes basic familiarity with agents and authorization while explaining cross-domain delegation and its additional cooperation mechanisms.
What you can do after
Evaluate cross-domain agent cooperation by checking promises, credentials, local policies, constraints, reputation, and recorded outcomes.
How it connects
Connects delegation with authorization policies, promises, verifiable credentials, reputation, auditability, and governance.

Related concepts

This explainer was generated by Manning's knowledge graph from the text of Authorization in Action. The section links point to the parts of the book it draws on.