designing-agent-teams

To design a multi-agent coding team, provide a project brief to generate an initial plan mapping models to specific agent roles, along with a starter AGENTS.md file for governance and orchestration rules.

The best way to assign models to roles in multi-agent orchestration is to evaluate cost, quality, and speed tradeoffs, then map appropriate models to specific agent responsibilities and record the assignments in an AGENTS.md file.

Yes, you can add new agents to an existing multi-agent team configuration by triggering an agent addition review, which evaluates the new role against current governance and orchestration rules before updating the plan.

You need to review your multi-agent team configuration when rebalancing roles to optimize cost, quality, and speed tradeoffs, or when onboarding new platforms that require updates to escalation paths and governance rules.

Escalation path management for multi-agent coding teams defines how tasks transfer between agents when thresholds are met, ensuring clear role-to-model assignments and orchestration rules are recorded in the AGENTS.md governance file.

To generate a multi-agent orchestration plan, you need to provide a project brief or your current team configuration, which is then used to craft agent roles, responsibilities, and optimal model-to-role mappings.