Chorus-AIDLC/Chorus
The Agent Harness for AI-Human Collaboration, inspired by the AI-DLC (AI-Driven Development Lifecycle)
What it solves
Chorus addresses the complexity of managing AI agents within a professional development lifecycle. It provides the necessary infrastructure to wrap around LLM agents to manage session lifecycles, task states, sub-agent orchestration, observability, and failure recovery, preventing the chaos of unmanaged autonomous agents.
How it works
Based on the AI-Driven Development Lifecycle (AI-DLC) methodology, the platform uses a "Reversed Conversation" philosophy where AI proposes and humans verify. It utilizes a permission matrix (5 resources × 3 actions) to grant specific capabilities to humans and agents. The system supports multiple agent backends (like Claude Code or Codex) and uses a daemon to turn local machines into resident agent runtimes, allowing for remote task dispatching and real-time monitoring.
Who it’s for
- Developers looking to integrate autonomous agents into their local workflows via a daemon.
- Project Managers who need to oversee AI-driven proposals, task DAGs, and requirements elaboration.
- Teams requiring a structured, permission-gated environment for human-AI collaboration.
Highlights
- Agent Harness: Manages session lifecycle, task state, and sub-agent orchestration.
- AI-DLC Workflow: Structured stages from Idea to Delivery with fine-grained permissions.
- Real-time Observability: Live terminal output, Kanban boards, and task dependency DAGs.
- Chorus Daemon: A resident runtime that connects local agent CLIs to a remote Chorus server.
- Visual Project Mapping: A graph view that wires ideas, proposals, documents, and tasks into a single tree.