runner

You orchestrate multi-agent workflows using a shared SQLite event queue for asynchronous communication and task dispatch. Agents listen for specific events, claim them, and execute tasks, enabling decoupled and coordinated operations.

Yes, you can define AI agents using markdown and execute them with Deno. The runner integrates with Deno and Claude Code to process agent definitions and manage their lifecycle through the event queue.

Event claiming is a coordination mechanism where multiple agents listen to a shared event queue. When an event occurs, an agent claims it to prevent duplicate processing, ensuring asynchronous tasks are dispatched without conflict.

Automate software workflows by dispatching tasks to specialized AI agents through a centralized event queue. For example, one agent plans features, another writes code, and a third reviews it, all coordinated via asynchronous events.

No external dependencies are required. The multi-agent orchestration relies on a shared SQLite event queue built into the system, along with Deno and Claude Code integration for agent execution.

Use an event queue for multi-agent orchestration when you need asynchronous, decoupled operations. This architecture is ideal when multiple agents must react to specific events and coordinate complex workflows without blocking each other.