tier-architect

To configure multi-model agent orchestration, you map orchestrator, specialist, and worker roles to specific model identifiers using a YAML tier list. This defines a tiered agent configuration ensuring planning, review, and execution stay aligned across your autonomous system.

A tiered architecture in autonomous agent systems is a configuration that assigns explicit parent-child relationships and specific roles, such as orchestrator, specialist, and worker, to different models. This structure organizes multi-model execution plans and decomposition workflows.

You set up a YAML tier list by defining entries with required model and role fields, optional parent references for hierarchy, and cold-start context rules. This valid YAML format is then used to assign responsibilities to specific models.

Yes, tier configuration supports cold-start context seeding by automatically instructing the system to read adjacent WIKI.md files. This refreshes operational context before executing decomposition workflows or fast discrete task runs.

Yes, you can use this tiered architecture setup for fast discrete task runs. The configuration applies to setting up multi-model execution plans, allowing specific worker models to handle discrete tasks efficiently within the autonomous agent system.