run

To start an autonomous development loop, you spawn the agent team and execute the run command. This initiates the orchestration state machine, driving work items from DISCOVER to DONE automatically.

Yes, you can resume the orchestration state machine after a session crash. The system performs cold resumes by reclaiming stranded work items and reconstructing runtimes from previously persisted state.

Yes, you need an initialized loop state and a valid behavior spec before running the autonomous agent team. The skill reads this state to spawn core teammates and manage phase transitions properly.

Autonomous software orchestration spawns always-on core teammates including a State Manager, PM, and Architect. It also spawns on-demand specialists like Builders and QA/UX reviewers as needed throughout the lifecycle.

The autonomous development loop handles phase transitions through the State Manager using explicit state transitions and non-blocking gates. It includes recovery handling to surface startup dashboards and execute recovery sweeps.

The best way to keep progress visible during autonomous agent orchestration is to use the built-in startup dashboard and recovery sweeps. These features surface work item statuses and reclaim stranded tasks automatically.