Select an agent's outward identity

Core meaning

Capability purpose: Ensure external recipients receive an intentional identity signal rather than one selected implicitly by a channel or runtime.
Meaningful outcome: An outward action has an owner-authorized identity mode specifying who is presented, whether AI involvement is disclosed, applicable scope, and escalation expectations.
Boundaries — includes: action or action-class scope, represented party, agent identity, disclosure state, channel context, recipient context, authorization source, expiry, revision, revocation, and escalation framing.
Boundaries — excludes: executing or sending the action, creating the agent identity, maintaining an AI representative, granting source access, and defining the content itself.
Terms and concepts: outward identity is the identity and disclosure signal presented to an external audience. It is separate from internal agent identity and runtime role.

Actor: The owner or an explicitly delegated identity-governance authority.
Command / intent: Select how an agent may present itself for a specific outward action or governed action class.
Current state: The acting agent, represented party, action, audience, channel, sensitivity, available identity modes, and authority can be determined.
Policies / invariants: No mode is assumed from technical convenience; undisclosed AI action requires the strongest applicable explicit authority; disclosed representative modes identify their relationship to the owner; expiry and revocation are enforceable; selection does not itself execute the action.
Transition: Evaluate identity, disclosure, audience, risk, and authority; record the selected mode or rejection for the bounded scope.
Result: An active, expired, revoked, or rejected outward-identity selection.
Events / effects: May authorize an outward action to proceed to its separate execution gate and supplies disclosure metadata to the delivery surface.
Unknowns: Universal disclosure rules, prohibited contexts, jurisdictional requirements, and standing-authority thresholds are not established.

The identity mode for an outward action must be explicit and attributable.
Selecting an outward identity must not grant permission to send, publish, transact, or disclose additional information.

Prefer clear AI disclosure and owner escalation paths; require action-specific approval for undisclosed representation unless an exact standing rule applies.

Legal, professional, relational, and channel-specific restrictions on undisclosed AI representation require dedicated profiles.

Statement	Semantic class	Evidence status	Source / note
Outward identity and AI disclosure are explicit owner-governed choices separate from agent identity, runtime role, and action execution.	rule/invariant	sourced	J07–J09.