agent-teams

Parallel multi-agent workflows are orchestrated by deploying specialized role-based agents, such as an Architect, Builder, and Validator, to execute independent modules simultaneously. This coordinated parallelization reduces total development time for complex projects.

Inter-agent messaging keeps parallel sessions in sync by utilizing a shared task list with dependency tracking. This mechanism allows multiple agents to coordinate their actions, communicate updates, and verify cross-agent progress throughout the project workflow.

Yes, multi-agent orchestration requires the CLAUDE_CODE_EXPERIMENTAL_AGENT_TEAMS environment variable to be set to 1 for activation. This experimental environment configuration must be enabled before instructing the orchestrator to initialize a team.

Yes, you can assign specialized roles like Architect, Builder, and Validator to different agents simultaneously. The skill deploys these role-based teammates to handle distinct aspects of a project, such as designing, implementing, and testing modules in parallel.

You should use parallelized agent teams instead of sequential task execution when building large features with independent modules. This approach solves the bottleneck of sequential execution by allowing simultaneous implementation and cross-agent verification.

The shared task list approach requires independent modules to be effectively parallelized for maximum benefit. Complex projects with heavy inter-module dependencies may limit the parallelization gains achievable through inter-agent messaging and coordinated workflows.