CHORUS and Chorus AIDLC: which one do you mean?

Two things in AI work share one name, and they are not even the same kind of thing.

CHORUS, in capitals, is a coordination protocol: six practices a small team writes down so that work shared with AI agents does not get lost at the handoffs. It comes from the book The Multi-Agent Organization. Chorus, from the GitHub organization Chorus-AIDLC, is open-source software: an “agent harness” you install, which hands tasks to AI coding agents, tracks what they do, and lets people verify what they propose.

One is a method. The other is a tool. You can use either without the other, and you can use them together.

A few words first. An AI agent is a program built on a language model that takes several steps toward a goal and uses tools along the way, such as reading files or running code. A coding agent is an agent that writes and changes software, such as Claude Code or OpenAI’s Codex. A harness is the software wrapped around agents that starts them, feeds them work, keeps track of their state and recovers when they fail.

What is Chorus AIDLC?

Chorus is an open-source project published on GitHub by an organization called Chorus-AIDLC (display name “Chorus AI-DLC”). The repository’s one-line description reads: “The Agent Harness for AI-Human Collaboration, inspired by the AI-DLC (AI-Driven Development Lifecycle).” Its website sums up the idea in four words: “AI proposes, humans verify.” (GitHub; chorus-ai.dev)

According to its README:

• You run it yourself. It is a web app you install and start with two commands, or deploy with Docker or on AWS.
• Agents plug in through MCP. Coding agents such as Claude Code, Codex, Kiro and OpenCode connect over MCP, the Model Context Protocol, an open standard that lets AI applications talk to outside tools and data.
• Work moves through fixed stages. An idea becomes a proposal (a requirements document plus a task DAG, a map of which tasks depend on which), agents execute the tasks, a person verifies, and the work is done.
• Each agent gets its own permissions, and a kanban board shows which agent is working on what.
• It is licensed AGPL-3.0, an open-source license. At the time of writing, the latest release was v0.19.0, published on 23 September 2026.

AI-DLC is not Chorus’s invention. It is a method described in a July 2025 post on the AWS DevOps blog by Raja SP, in which AI “creates a plan, asks clarifying questions to seek context, and implements solutions only after receiving human validation” (AWS). The Chorus project says it is inspired by that method. Neither its GitHub page nor its website presents it as an AWS product.

What is CHORUS, the protocol?

CHORUS is a way of working for teams of about 2 to 15 people who share work with AI agents. Its job is to prevent the most common failure when a team adds agents: work gets slower, because nobody wrote down who owns a task at the boundary, what context travels with it, and who checks the result. The site calls that the coordination gap.

It has six steps. The names are fixed; the lines are plain-words versions.

1. Charter. Describe every role, human or agent, in seven fields: purpose, inputs, outputs, authority ceiling (how far it may go on its own), success criteria, review requirement and escalation.
2. Handoff. When a task moves to the next worker, send a short context packet (goal, constraints, inputs, where they came from, acceptance criteria, deadline) so the receiver never has to guess.
3. Orchestrate. Decide the shape of the flow: steps in order, work split across parallel agents, one named place where the results merge, and a named person to escalate to.
4. Review. Output passes a gate only when it meets written acceptance criteria, not because it sounded right.
5. Update. What reviews find flows back into prompts, tools, procedures and charters, so the next run is better.
6. Sync. One shared surface shows who owns what and what is blocked, read in a short weekly check-in.

CHORUS comes from The Multi-Agent Organization, a book by the author of this site, which is available now. You do not need the book to use it. The framework page and What is CHORUS? cover the six steps.

The book is blunt about software: “Order matters: protocol first, tool second, always.” Its reasoning is that “A charter is a decision about authority, written down.” You can run all six steps with a shared document and a chat thread. Tools make the steps faster and harder to skip. They do not make them exist.

CHORUS vs Chorus: side by side

	CHORUS	Chorus (Chorus-AIDLC)
What it is	A coordination protocol: six practices	Open-source software: an agent harness
Who publishes it	Len P. van der Hof, in The Multi-Agent Organization	The GitHub organization Chorus-AIDLC
How it is written	All capitals; each letter is a step	Capital C only; “AIDLC” is part of the organization’s name
What you get	Written charters, handoff packets, review gates, a weekly sync	A web app, a task pipeline, agent permissions, a kanban board
Where it runs	Anywhere: a shared doc, a task board, any software	On your own machine or server
Who it is for	Small teams sharing any kind of work with agents	Software teams working with coding agents
Core idea	Nothing crosses a boundary without its context; nothing ships without a gate	”AI proposes, humans verify”
Do you install it?	No	Yes

Are there other things called Chorus?

Yes. Two you may meet near AI and sales work:

• Chorus by protolabs42 (chorusprotocol.dev) calls itself a “Self-hosted shared workspace for agent teams”: shared memory, signals and file handoff for AI agents.
• Chorus by ZoomInfo (zoominfo.com) is a conversation-intelligence product that captures and analyzes sales teams’ calls, meetings and emails.

Neither is related to the CHORUS protocol or to Chorus-AIDLC.

How do you tell which one a document means?

Five checks, in order. The first one that fires usually settles it.

1. Is there an install step? An install command, a Docker file, an API key, a login screen: that is software. CHORUS has nothing to install.
2. Do the six words appear together? Charter, Handoff, Orchestrate, Review, Update, Sync: that is the protocol.
3. Do these words appear? “Harness”, “AI-DLC”, “proposal”, “task DAG”, “daemon”, “verify”: that is Chorus-AIDLC.
4. Where does the link go? github.com/Chorus-AIDLC or chorus-ai.dev is the harness. lenvanderhof.com or the book is the protocol. chorusprotocol.dev is the protolabs42 workspace. zoominfo.com or chorus.ai is the sales product.
5. Still unsure? Ask one question. “Is this something we install, or a way we agree to work?”

Can you use both?

Yes, and doing so shows the difference clearly. Imagine a hypothetical four-person software company: two developers, a product lead and a founder, plus three coding agents. They install Chorus-AIDLC so the agents can pick up tasks and the founder can verify proposals in one place.

The software gives them a pipeline. It does not decide how they work. CHORUS asks the questions no install answers:

• Charter. Which agent may draft proposals, and which may never merge to the main branch? Who may approve a proposal?
• Handoff. What must the product lead write into an idea before an agent picks it up?
• Review. What written line does the person at the verify stage check against?
• Update. When a proposal is rejected, where does the lesson go so the next proposal is better?
• Sync. Which board do the four people read on Monday, and for how long?

Some of the software’s features line up with CHORUS steps. The proposal and task DAG are a form of Orchestrate, the verify stage is a Review gate, the kanban board serves Sync, and agent permissions carry part of a Charter. That mapping is this page’s reading of the README, not a claim either project makes. The content of a handoff and the habit of updating stay with the team, whatever tool it uses.

Why does the name clash matter?

Mostly it costs time: a search for one lands you on the other. It costs more when the word travels into a decision. Imagine a hypothetical procurement form that says “supplier must support CHORUS”. One reader hears “install this open-source harness”. Another hears “write charters and review gates”. Both tick the box. They mean different things.

The fix is cheap. Write the protocol in capitals with its six words the first time you use it. Link the software to its repository the first time you mention it.

Try this today (10 minutes)

Search your team’s documents and chat for “chorus”. Run the five checks on every hit and write one word next to it: protocol, harness, workspace, sales or unclear. For each “unclear”, add a link or the six words so the next reader does not have to guess.

If the hits say “protocol” but nobody can show you a written charter or a handoff packet, you have the name without the practice. Start with one charter for the agent your team trusts most.

Cite this:CHORUS and Chorus AIDLC: which one do you mean?Len P. van der Hof. https://lenvanderhof.com/en/blog/chorus-name-collision/ · Published 1 October 2026.