Chorus-AIDLC/Chorus

The Agent Harness for AI-Human Collaboration, inspired by the AI-DLC (AI-Driven Development Lifecycle)

Chorus – Harness for Coding Agents

One‑line description: A platform that orchestrates multiple AI coding agents together with human reviewers, turning ideas into shipped software.

What it does

  • Implements the AI‑DLC (AI‑Driven Development Lifecycle) workflow: Idea → Proposal → Task DAG → Execute → Verify → Done.
  • Provides a web UI (Next.js 15, React 19) where you can create ideas, view live project graphs, Kanban boards, and monitor agent activity.
  • Manages session lifecycle, task state, sub‑agent orchestration, observability, and failure recovery.
  • Offers fine‑grained permission matrix (5 stages × 3 roles) that can be assigned to humans, agents, or both.
  • Supports a variety of agent back‑ends via the MCP SDK (Claude Code, Codex, TRAE CLI, Kiro, Pi, Hermes, etc.).
  • Includes a daemon that runs on a machine, picks up assigned tasks, and reports back without manual terminal work.
  • Self‑hosting options: single‑command install with embedded PostgreSQL (PGlite) or full production deployment via Docker Compose, external PostgreSQL 17, optional LAN HTTPS and Cloudflare Tunnel.

Key components

Component Tech
Front‑end Next.js 15, React 19, Tailwind, shadcn/ui
Backend TypeScript 5, Prisma 7, PostgreSQL 17, optional Redis
Agent integration MCP SDK (HTTP streamable transport)
Auth OIDC + PKCE, API keys, SuperAdmin
CLI @chorus-aidlc/chorus (npm global) – install, run daemon, upgrade plugins

Typical workflow

  1. Human creates an Idea.
  2. An AI agent drafts a Proposal (PRD + task DAG).
  3. Tasks are assigned; agents execute code while the UI shows live status.
  4. Humans verify results; once approved the workflow ends.

Getting started (quick)

npm install -g @chorus-aidlc/chorus@0.21.0
chorus   # launches UI at http://localhost:8637 (admin@chorus.local / chorus)
  • No external DB needed for the demo; an embedded PostgreSQL runs automatically.
  • For production: follow the Deploy & self‑host guide (Docker Compose, external PostgreSQL, Cloudflare Tunnel).

Extending / connecting agents

  • Use the in‑app wizard or the docs to generate an API key (cho_…).
  • Supported agents include Claude Code, Codex, TRAE CLI, Kiro, Pi, Hermes, OpenCode, OpenClaw, and any MCP‑compatible tool.
  • chorus daemon runs on a remote machine; it watches assigned directories and executes tasks automatically.

Who might use it

  • Product teams that want AI‑generated specs and code with human oversight.
  • DevOps / platform engineers looking to run AI agents as part of CI/CD pipelines.
  • Researchers experimenting with multi‑agent coordination and human‑in‑the‑loop workflows.

License

  • AGPL‑3.0 (see LICENSE.txt).

Documentation: https://doc.chorus-ai.dev
Discord community: https://discord.gg/SwcCMaMmR

Related

  • Project
    chorus-codes/chorus

    A local orchestration tool that uses multiple AI subscriptions to provide parallel, multi-vendor code reviews to catch bugs and biases that a single AI model would miss.

  • Project
    awslabs/aidlc-workflows

    AI-DLC is a structured, harness-neutral framework that turns AI coding assistants into verifiable, audited software delivery workflows with human approval gates and multi-stage lifecycle management.

  • Project
    affaan-m/ECC

    ECC is an open‑source framework that adds a full engineering workflow (plan‑test‑implement‑review‑verify‑remember‑improve) to AI coding agents like Claude Code, Codex, and Kimi Code. It ships 71 agents, 311 skills, security scanning (AgentShield), and a guided installer (`npx ecc-universal@2.2.3 setup`). MIT‑licensed, with optional paid Pro features for private‑repo use.

  • Project
    cyrusagents/cyrus

    Cyrus is an AI agent orchestrator that monitors issues in Linear, GitHub, GitLab, or Slack and uses AI coding agents to automatically resolve them in isolated Git worktrees.

  • Project
    Arch1eSUN/Arcgentic

    A harness engineering layer for Codex and Claude Code that transforms ad-hoc AI prompting into a gated, role-based engineering workflow with audits and stop states.