The AI Memory Package Wizard keeps Base mode as the simplest path and exposes advanced workload modes only when the user selects them. One workload class deterministically activates its required surfaces, fields, files, and validators.
Nine-step flow
- Choose a setup mode and preset.
- Name the package and boundary.
- Declare collaboration, architecture, the advanced workload, authority root, roster, identity layers, and memory architecture.
- Assign operations, tasks, ownership, coordinator, delivery, conflict, and recovery policy.
- Inventory credential file metadata without secret values.
- Configure protocols, source authority, governance, review, and evidence.
- Configure active memory and any selected durable-memory pointer or endpoint.
- Record the receiver, first bounded task, and success evidence.
- Review, validate, and export.
Stable setup routes
#setup-multi-agent-codebase#setup-agent-swarm#setup-database-npcs#setup-multi-agent-MATM#setup-multi-agent-codebase-MATM#setup-database-npcs-MATM
Update URL contract
The generated .uai/readme.human, AGENTS.md, and .uai/agent-pickup.uai name the exact wizard setup URL used to create the package. That URL updates package configuration; a MATM update_url names the authenticated durable-memory update route. They are not interchangeable.
Support boundary
Agent runtimes execute. UAIX records portable configuration, logical identity, authority boundaries, routing references, reviewed evidence, and offline continuity. Runtime systems retain authentication, credential verification, leases, queues, transactions, tool execution, database writes, and delivery readback. A missing-authority request is review-pending only and applies no grants.
Schemas and validation
- Advanced Multi-Agent Workload schema
- Reference workload example
- UAI-1 profile registry
- Swarm Readiness checklist