Optionaloptions: AgentFeedbackCoordinatorOptionsBuilds the standard editable plan-approval AgentResponseForm for Plan Mode.
Creates and saves a persistent MJ: AI Agent Requests row for a Chat or Plan step,
returning the new request's ID when saved or null on failure.
Who saves the row. MJAIAgentRequestEntityServer.Validate keeps requests private: a
non-Owner-type user may save a request only for themselves, and Owner-type users (the
scheduler's system user among them) are exempt. That rule guards direct saves, such as a
GraphQL create where the caller types in RequestForUserID, and this method leaves it
untouched. The assignee here comes from one of two places, held to different standards:
AgentWakeTriggerService saves a wake routine for another user). Saving it as
the caller would fail validation and lose the request: no inbox row, no resume.ExecuteAgentParams.assignmentStrategy): held to the rule a direct save
meets. A non-Owner caller who names someone else gets the request assigned to themselves
(logged), and the row is saved as the caller, so the entity rule still backs this up.
The reads that choose the assignee (list members, open counts) run as the caller either way.Inspects an AgentResponseForm to determine the request type name.
Returns 'Approval' when the form is a single two-option approve/reject question,
otherwise 'Information'.
Resolves the assignee user ID for a feedback request, including async List
strategies (RoundRobin, LeastBusy, Random).
OptionalagentRun: MJAIAgentRunEntityExtendedOptionalprovider: IMetadataProviderWalks the assignment strategy resolution chain (bottom-up, first-non-null wins):
ExecuteAgentParams.assignmentStrategy (per-invocation override)AssignmentStrategyAssignmentStrategy (walks up ParentID tree)DefaultAssignmentStrategynullWalks up the agent's category hierarchy looking for a configured AssignmentStrategy.
Resolves the MJ: AI Agent Request Types ID for a Chat or Plan step decision:
"Approval""Information""Information"Request types come from the AI engine's cached MJ: AI Agent Request Types rows
(AIEngineBase.ResolveAgentRequestTypeID), the same source the agent actions use.
OptionalcontextUser: UserInfoSynchronously resolves non-List assignment strategies (RunUser, AgentOwner,
SpecificUser, SharedInbox, or null fallback).
OptionalagentRun: MJAIAgentRunEntityExtended
Coordinates creation of persistent
MJ: AI Agent Requestsrows, response-form classification, hierarchical assignment strategy resolution, andList-based (RoundRobin,LeastBusy,Random) user assignment.