Chorus and open-harness
These two projects are competitors: A is a more established Agent Harness for AI-Human Collaboration, while B is a less mature, code-first SDK to build powerful AI agents, both aiming to provide frameworks for developing AI agents.
About Chorus
Chorus-AIDLC/Chorus
The Agent Harness for AI-Human Collaboration, inspired by the AI-DLC (AI-Driven Development Lifecycle)
Manages multi-agent LLM workflows through persistent sessions, strict task state machines, and context continuity across restarts—enabling PM, Developer, and Admin agents to collaborate on projects with real-time observability via Kanban boards and task DAGs. Built on the AI-DLC methodology with a "reversed conversation" pattern where AI proposes and humans verify; integrates with Claude Code Agent Teams via MCP tools for sub-agent orchestration and automatic lifecycle management (session creation, heartbeats, task handoff).
About open-harness
MaxGfeller/open-harness
A code-first, composable SDK to build powerful AI agents
It is built on Vercel's AI SDK and offers core packages for agent execution, conversation management, middleware, and tool integration, alongside React and Vue hooks for streaming chat UIs. The SDK features composable middleware for turn tracking, context window compaction, and retry logic, enabling multi-turn conversations and custom tool creation for tasks like filesystem access or bash execution. It targets AI models via providers like `@ai-sdk/openai` and supports nested subagents and the Model Context Protocol.
Related comparisons
Scores updated daily from GitHub, PyPI, and npm data. How scores work