CCCC: Coordinate Your Coding Agents Like a Group Chat

Introduction

CCCC, short for "Coordinate your coding agents like a group chat," is a powerful, production-minded orchestrator for 24/7 AI agent workflows. It brings together various coding agents, such as Claude Code, Codex, and ChatGPT Web, into one persistent and coordinated team. With a single pip install and zero additional infrastructure, CCCC simplifies the management of multi-agent systems across different runtimes, machines, and trusted working groups, moving beyond disparate terminal sessions.

This Rust-based tool provides essential features like read receipts, delivery tracking, and the ability to perform remote operations directly from your phone, ensuring seamless collaboration and oversight.

Why Use CCCC and Its Advantages

When orchestrating multiple coding agents, it's crucial to track task assignments, message delivery status, and maintain progress persistently across restarts. CCCC addresses these challenges by providing a centralized platform for management and operation, accessible via web or instant messaging, even when you're away from your terminal.

CCCC operates your agent collective as a persistent, coordinated system, offering several key advantages:

Persistent Coordination: Message history is recorded in an append-only ledger, with tasks and shared contexts also persistently stored.
Visible Delivery Facts: It meticulously tracks routing, storage, runtime delivery, read receipts, and replies independently, ensuring that "sent" doesn't equate to "confirmed."
Unified Control Plane: A single daemon state is shared across the Web UI, CLI, MCP (Multi-Agent Control Protocol), and IM bridges, preventing state fragmentation.
Multi-Runtime Orchestration: Seamlessly mix and match major runtimes like Claude Code, Codex CLI, ChatGPT Web, and Grok Build within a single group.
CCCC Connect for Instance Linking: Supports discovery within the same account, authorized group connections between members, and account-free direct connections, maintaining instance state and permission boundaries.
Local-First Operation: Get started with a single installation command, keeping runtime state in CCCC_HOME and extending to remote monitoring only when needed.

CCCC is designed for simplicity, requiring no separate database, message broker, or Docker setup for its core functionality.

Installation

CCCC is a single native Rust implementation. The website installers are recommended for ease of use.

# macOS / Linux (Recommended)
curl -fsSL https://chesterra.github.io/cccc/install.sh | sh

# Windows CMD or PowerShell (Recommended)
powershell.exe -NoProfile -ExecutionPolicy Bypass -Command "[Net.ServicePointManager]::SecurityProtocol = [Net.ServicePointManager]::SecurityProtocol -bor [Net.SecurityProtocolType]::Tls12; Invoke-RestMethod 'https://chesterra.github.io/cccc/install.ps1' | Invoke-Expression"

# pip compatible native platform wheel
python -m pip install -U "cccc-pair>=0.4.36"

Examples

To get started, ensure your agent CLIs are installed and logged in. Here's an example using Claude Code and Codex CLI to create a multi-agent group:

cd /path/to/your/repo
cccc attach .                              # Attach the directory as a scope
cccc setup --runtime claude                # Prepare only the runtimes used in this example
cccc setup --runtime codex
cccc actor add foreman --runtime claude    # The first actor becomes the foreman
cccc actor add implementer --runtime codex # Add a peer
cccc group start                           # Start all actors
cccc send "Review the repository and suggest the first safe task." --to foreman
cccc tracked-send "Take on the first concrete task and reply with verification evidence." \
  --to implementer \
  --title "First Concrete Task" \
  --outcome "Changes and verification evidence reported"

With these steps, two agents will be collaborating within a persistent group, complete with a full message history, delivery tracking, and a web dashboard. The daemon handles delivery and coordination, while runtime state remains outside your repository in CCCC_HOME.

You should see both actors running in the Web UI at http://127.0.0.1:8848, with the foreman's reply appearing in the chat and delivery/read statuses displayed for tracked requests.