amoa-two-phase-mode

Multi-agent implementation projects can be coordinated by breaking user goals into modules, creating GitHub issues, and registering AI and human agents for assigned tasks. This plan-then-execute workflow enforces periodic polling and verification loops to monitor progress audibly.

Instruction verification in multi-agent orchestration is a mandatory pre-start protocol that runs checks before execution begins. It prevents misunderstandings and unsafe changes by enforcing a 5-step update verification flow for all implementer actions.

You can sync GitHub issues with AI agents by using the gh command line tool to create and track modules from a YAML state file. This process assigns registered human or AI implementers to specific GitHub issues and monitors their progress.

Yes, verification loops can be enforced before task exit using a phase-aware stop hook. This hook blocks completion until plan and orchestration exit criteria, including multi-loop verification and mandatory polling questions, are fully satisfied.

Yes, the gh CLI is a required dependency for managing GitHub issues during orchestration. It is used alongside pyyaml to maintain YAML state files and sync tasks across your registered agents.

Polling limitations include mandatory 10–15 minute intervals where six mandatory questions must be answered, and implementer issues are strictly triaged using a 3-choice rule to explain, rewrite, or escalate, which may slow down rapid iterations.