Shared Agent State
Examine how cooperating agents should coordinate task facts without blurring ownership or access control.
Original scenario-based practice challenge for Agentic Architect Lab. This is not an official certification question.
Scenario
Several agents need the same task facts, but not the same edit rights. How should that coordination work?
A proposal workflow uses three cooperating agents: one gathers customer requirements, one computes pricing, and one checks legal constraints. They all need access to shared task facts such as customer tier, approved discount ceiling, and jurisdiction, but pricing should not edit legal approvals and the legal agent should not overwrite commercial assumptions. The team also wants a clear audit trail when one agent updates shared information.
Which design best supports cooperation without turning shared context into an unbounded or ambiguous memory channel?