cdo

What problem does it solve?

CDO provides a safe, structured way to orchestrate multiple AI agents to complete complex tasks that require delegation, parallel execution, and cross-system coordination. It prevents uncoordinated agent runs and ensures results are produced only when explicitly requested by the user and allowed by the environment.

Core Features & Use Cases

• Agent orchestration: classify tasks into a pipeline of agents (e.g., guardian, oracle, nexus, screen) and run them in the correct order with support for parallel waves.
• Agent bus communication: agents read and write results from a shared bus (.claude/agent-bus) and can escalate issues via @mentions to trigger additional waves.
• Phase-based pipelines: supports SEQUENTIAL, PARALLEL, and HYBRID modes with clear gate checks and deployment-ready templates.
• Safeguards and escalation: built-in blocks for blockers, safety prompts, and escalation when tasks fall outside an agent’s competence.

Quick Start

Enable CDO only when the user explicitly requests orchestration or parallel agent work and the environment allows subagents.